Who can help? #551
Replies: 109 comments 78 replies
-
|
Not here to offer help on maintaining per se (I'm not qualified), but just a few heads up in general if that's okay:
Anyway best of luck on finding maintainers for the repo! I really like this OSP and hope to see it overtake the biggest companies on the planet! <3 |
Beta Was this translation helpful? Give feedback.
-
|
Maybe he needs to add the Discussion feature to this project or somewhere. I know that's a feature of GitHub, but I don't know if it's for Organizations only or if someone can do it for an individual repository. |
Beta Was this translation helpful? Give feedback.
-
|
I work as Head of Technical Operations for an AI legal contract review company. I've been developing for 10+ years though admittedly the last time I wrote something myself as opposed to using Claude / Cursor / co-pilot is probably 2 years ago, but you just add "don't make mistakes" right... I've got a windows machine with a RTX 3070 8Gb and a Macbook M4 with 24Gb RAM that I can do testing on. Realistically I could offer an hour or 2 a day right now and I haven't even set the project up for myself yet, but just about to get started and I absolutely LOVE the concept. |
Beta Was this translation helpful? Give feedback.
-
|
Hi pewds, I just finished my CS degree, and my most relevant experience is my bachelor thesis. I took an existing conversational LLM assistant and rebuilt its understanding/response layer around an LLM running through Ollama, with a smaller model kept as a deterministic fallback for when the LLM was down or confused. So I spent a lot of time deciding when a local model was more useful, and handling "I can't do that" gracefully. I don't have experience as an open source maintainer, but a lot of work went into getting the prototype running again (dependency drift, dead library APIs, bugs and such), which got me comfortable with PRs, issues, and code reviews. On the side I built and still maintain my solo project GuitarDex, a retro styled React/Vite PWA, and I've done GUI + debugging work on a Rust data platform. Time-wise I am mostly free the next few months ahead, so I can realistically do around 10 hrs/week. I think I'd be a nice fit as the stack lines up well with what I've already used: Ollama, Python backend, React PWA frontend, and I'm on Linux daily. So where I'd help most is testing on Linux, deduping issues, frontend/UI cleanup, and reviewing PRs in the Python/React parts I'm familiar with. I've gotten the project running on my Linux :) Thanks for releasing this open in the first place, very exciting project! |
Beta Was this translation helpful? Give feedback.
-
|
Hello, I currently work full-time as a software engineer and have about two years of professional experience, along with a degree in Computer Science. Most of my work is backend development in Java, but I've worked with a variety of languages and technologies over the years. I don't have any open-source maintainer or reviewer experience yet, but reviewing code is a regular part of my job. I often review pull requests and give feedback, even though I'm not the final person deciding whether something gets merged. I can realistically commit at least 16 hours a week to the project and I'm located on the East Coast of the US. For testing, I have a Linux development machine as well as a separate Windows 11 VM. I also have access to both AMD and NVIDIA GPUs, so I can help test across different operating systems and hardware configurations when needed. Thanks for making Odysseus! |
Beta Was this translation helpful? Give feedback.
-
|
Hi PewdiePie, I have watched your video since... forever. I absolutely love the idea of making AI local. I am a rising Senior in CS. I am fairly proficient with Python, and C++. I have a Windows 11 with a Linux partition that I was inspired to set aside after watching your Linux build... I still need to set that up. Most of my coding I do is on a macOS. I am not certain how qualified I am ; I have used AI to build websocket throttling apps, or a windows monitor that auto quits and logs my gaming so I can review how much I play (not that it prevents me from playing because I can still force quit the python script but sometimes I forget and it works and I go to sleep). I have done a lot of algorithms work; Mostly studying greedy algorithms, BFS, DFS, BST, algorithms etc. Not sure how much use that is but that is what I am most studied in outside of formal Python, CSS, Computer Architecture. I am excited about this project and would be happy to help. I have some construction management background, about 5 yrs, but in terms of CS I am fairly new (still in school). I would love to help in any way possible, and though I have not contributed formally to a repo outside of school projects yet, I have an internship at an AI startup so I will be getting github experience there as well. Odysseus looks awesome and I think it will make me finally start my Linux partition. Not sure if I can help a ton given my lack of formal experience but I would love to help in any way possible. |
Beta Was this translation helpful? Give feedback.
-
|
Hey PewdiePie, I have been doing coding ever since my high school and I think I can really help with Odysseus. I am well equiped with knowledge around python, node & rust ecosystem and everything around it. I have been a maintainer in open source repos: https://github.com/code100x/cms & https://github.com/code100x/daily-code I have about 25hrs per week of free time to work on this project. I love building stuff and ever since I have been into cs I have building only. I have worked with multiple open source and closed source companies and can understand requirements from pure english and convert it into code that makes sense for business. Thanks, sargampoudel100@gmail.com |
Beta Was this translation helpful? Give feedback.
-
|
Been programming since I was around 10 years old, am 18 now, and I would be willing to lend a hand with PR reviews whenever I have the time for it. I really look forward to seeing the project grow and as someone who hasn't been very keen on using AI for everything, I think this project could be a start to using AI to organize myself a little better. Most of my recent time spent programming have gone to more theoretical aspects, (type theory, functional programming, compilers), so I am not a front-end kind of guy, regardless I know my way around most mainstream languages and have worked with large code-bases before. I was team lead for open-contribution studio, working with game dev, with 10+ people in the team. Most of my expertise lies with lower level languages, ive used Rust a lot the past year (looking into the rustc repo), but I am of course well accustomed to Python, as well as C#, Java same same etc. My biggest strength lies in software architecture rather than development and iteration. I don't have too much time on my hands, but I am able to communicate always. Though i will be freer now with summer break starting in roughly a week. If i were to give an estimate i would say around upwards to 8h/week of dedicated time but counting time being involved in anything to do with the project i would say maybe twice that. I don't feel the need to try and argue my worth too much for this project, if help is wanted I am down to participate! I am not looking for a big role, not because i lack the ability but because I don't see myself being able to prioritize this project as much as probably many others who would fit better in those roles. I just wish to help where I can, and I think I do so best by being close to important discussions and decisions regarding the project, where I can provide insight. I am a little scared of under-qualified people with a bit too much pride in their abilities being given big roles (have seen many projects suffer from this), so I urge you to be very precarious of whom gets the big roles. I can't stress enough how refreshing it is to see someone with your level of influence, hold such good morals regarding the digital world. I'm looking forward to seeing where all this goes. 😛 ! |
Beta Was this translation helpful? Give feedback.
-
|
Hey, pewdiepie love the new tech congrats on your release. We've raised over 156 Solana so far ($12,480 USD) for you to help the longevity so you don't have to continue building on your own. You can use it to hire programmers/developers or even help with licensing. The money is on PumpFun they have a public IPO its a billion dollar company google it before you claim so you can verify on your own that the community is not a scam. It's completely yours and will sit there forever as time goes on we will try to raise more for you. Odysseus will grow to be household ai I can see it now! We're so proud to see what you have in store for the future. Odysseus PumpFun Community https://pump.fun/coin/GmshYgy1nqUxw9ooza8UUnTw1rxDKM5t7HP4KSYVpump |
Beta Was this translation helpful? Give feedback.
-
|
This sort of "claim your unclaimed crypto" is a well known scam especially in the social media space. Skepticism is duly required upon seeing this in a "who can help" thread. |
Beta Was this translation helpful? Give feedback.
-
|
"Hey one of the most popular people in the entire world, multi-Millionaire retired before midlife, we raised 12 grand to associate your passion project with a scam" My brother in Christ, what the actual fuck are you doing? Also calling it pump fun is... very funny. |
Beta Was this translation helpful? Give feedback.
-
|
Hey, I’m interested in helping maintain/review. I’m Bandonker. I’ve built a lot of personal local first desktop, AI, tooling, and game projects, mostly with agentic coding workflows. I’m comfortable working on implementation, debugging, testing, packaging, and runtime validation. I excel in fast-paced environments, but I also care a lot about proving the real user flow before calling something done. Relevant experience:
For Odysseus, PR #603 is a good example of what I can contribute. I added a scoped Windows/Tauri wrapper, portable packaging, docs, CI artifact workflow, screenshot evidence, and runtime validation. I can put in around 20 hours a week. I can help with Windows/runtime validation, PR review, issue triage, docs/setup cleanup, scoped fixes, and checking that user-facing flows actually work before they merge. I’m very familiar with agentic tooling, but I review and verify all outputs instead of treating them as automatically one and done. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, First off, congratulations on the strong attention the project is getting! It’s always a good sign when a repo starts attracting this level of contributions and community activity, I wish you good luck with this I am a professional software engineer with 6+ years of experience working the full development lifecycle including feature work, bug fixing, code reviews, and manual testing. I also triage issues, clarify requirements with stakeholders, and contribute to longer-term architectural design and technical decision-making. I’m comfortable reviewing PRs for correctness, scope, and regressions, and I can help with testing and general stability checks. I also have experience working across evolving codebases where maintaining consistency and avoiding unintended side effects is important, this includes framework migrations. I have a lot of experience reviewing AI-generated code for quality. I've been looking through the project files this evening and I'm already interested in where this is going. I don’t have prior formal OSS maintainer experience, but I operate in a review-heavy engineering environment and can apply that same discipline here. Thanks |
Beta Was this translation helpful? Give feedback.
-
|
Hello, I'd be interested in taking on a maintainer/reviewer role. I currently work at a cloud and Generative AI contracting agency, building softwares very similar to this project with a largely identical stacks. I can contribute through code reviews, issue triage, architecture discussions, and implementation improvements. As this is essentially my day-to-day work, I can realistically review 2–3 PRs and issues per day, above all i think quality is very important. I can also contribute from time to time as i am very enthusiestic about local AI and private AI solutions. Probably the most important i bring to the table is industry knowledge and enthusiasm for the vision. One example of my contributions is PR #120 and #123 . About me: I have a BSc in Computer Science, 4+ years of professional experience in this field, I'm a big open-source enthusiast, and I've been following your channel for over 10 years. I also enjoy bouldering. I really feel passionate about your current videos about data privacy and opensource. I also regularly attend hackathons. In my freetime i also spend time on local AI hobby projects such as automatic ebook translation using locally deployed AI, AI training models, building Graph based vector embeded knowledgebase. Relevant experience: Apex Lab (Nov 2025 – Present) – Cloud & AI Engineering Contractor AWS architecture, infrastructure audits, cloud migrations, and Generative AI solutions Robert Bosch Kft. (2022 – 2025) – Cloud Developer & Business Application Developer Worked on Metron, a KPI and engineering analytics platform similar to DevLake More information about me can be found on my portfolio (https://varghacsongor.hu/about/) and GitHub profile. Thanks for your consideration. Brofist from Hungary |
Beta Was this translation helpful? Give feedback.
-
|
Hello I've actually been working on a similar project myself, mythosorgin.com Its barebones right now, but has tauri and electron build so it is an app you can run and download easily with an installer, i should be able to do the same for this, and fix other issues that I faced while creating mine too. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, |
Beta Was this translation helpful? Give feedback.
-
|
Hey! im interested but a few thigs Im bandit, software engineering student, i run Arch Linux with an RTX 3060, so Linux + GPU testing is something I can actually do properly, not just in a VM, i also run ollama locally so I can test the model serving side too Honest disclaimer: I'm in my last weeks of semester right now, so I can't promise 20hrs/week, maybe 1-2 hours daily, tops 8 hours by week, im sorry on that department I don't have prior OSS maintainer experience, but I'm careful with user-facing changes and i wont merge things i dont fully understand and ill do everything to undertsand the full repo Let me know if that's useful, if not have a good one (pewds and readers alike!) |
Beta Was this translation helpful? Give feedback.
-
|
I would also love to help whenever needed. Intro, relevant experience
Any prior open-source maintainer/reviewer experience
How much time you realistically have
Why you think you’d be a good fit
|
Beta Was this translation helpful? Give feedback.
-
|
Hi, I’d be interested in helping, especially with testing, reproducibility, documentation, and reviewing user-facing behavior. My background is in software testing and requirements management. I worked for many years in software development environments, including test management and requirements engineering, with a strong focus on testability, reproducible defects, regression risk, and clear acceptance criteria. I’m currently running Odysseus locally on macOS and actively testing real-world workflows around local models, LM Studio / OpenAI-compatible APIs, MCP/tools, email, calendar, notes/tasks, browser/tool routing, and local patches. I’ve also been comparing my local changes against main, keeping patch files, and checking whether fixes in main make local patches obsolete. Where I think I could be useful:
I’m retired now, so I have some flexibility, but I would prefer to contribute in focused sessions rather than promise full-time availability. I’m especially interested in helping keep Odysseus stable and understandable as it grows, not just adding more features quickly. Happy to help if this kind of testing/review/documentation support would be useful. |
Beta Was this translation helpful? Give feedback.
-
|
How I currently run Odysseus: I’m running Odysseus as part of a small local AI/home-lab setup that I call my “Tafelrunde” (“Round Table”). The idea is to use Odysseus not only as a chat UI, but as a local agent/workspace that connects models, tools, documents, notes, email/calendar workflows, and local infrastructure. My current setup looks roughly like this:
The practical goal is to test Odysseus in a realistic local multi-machine environment:
In practice, I often test things like:
So my contribution would mostly be from the perspective of a careful real-world user, tester, and requirements/reproducibility reviewer. I can help identify where user-facing behavior breaks down, document the exact steps, compare expected vs. actual behavior, and help turn vague issues into actionable bug reports or test cases. |
Beta Was this translation helpful? Give feedback.
-
|
Hi Felix, I’d be interested in helping, but in a fairly narrow area. My strongest fit is provenance / replayability around agent and tool execution: what decisions are made at runtime, what tools are allowed or actually used, how model/provider choices are surfaced, and whether user-facing behavior can be reviewed later without relying on chat history, memory, or guesswork. I’m not looking for broad maintainer responsibility. I think I’d be more useful doing focused review passes on PRs, issues, or runtime paths where agent behavior, tool execution, auth/secrets boundaries, model routing, or recoverability matter. I can start externally with a small pass over one concrete area and report back with actionable findings: what seems observable, what seems missing, what looks risky, and what I could or could not validate. I’ll be clear about the difference between static review, local testing, and anything that would need a different setup or hardware. I’ve been doing public hands-on R&D around traceable AI workflows and runtime observability, so this is a real area of interest for me, not just a vague “happy to help.” But I’d rather start from Odysseus’ actual shape and see whether the output is useful. I work best with bounded requests: point me at a PR, issue, or runtime path, and I’ll come back with findings rather than trying to take over a broad area. For transparency, I also kept a short public working notes for myself here so the offer does not live only as a GitHub comment: And for transparency here is the viewer I use to progress with traceable AI/human provenance: https://tiinex.dev/ (it's client side only) |
Beta Was this translation helpful? Give feedback.
-
|
Welcome to the thread, Olle! Your focus on runtime provenance and
deterministic execution tracing is exactly what an agentic framework like
`odysseus` needs to survive production scaling.
I see a really powerful bridge between our respective domains here. While
you are mapping out the high-level decision paths, tool routing policies,
and auditability metrics, I can focus on securing the low-level execution
boundaries directly beneath them.
For instance, when an agent triggers tool execution loops, my background
stack (`renorm-native`) can handle the hardware telemetry
validation—safeguarding the runtime against uncoalesced memory
fragmentation or runaway step allocations. If your tracking systems detect
a risky state or an unmapped provider choice, our low-level loop guards can
intercept and stabilize the execution graph before the OS layer panics.
If Felix throws us a concrete runtime path or an active PR to audit, your
pass can validate *what* decisions the agent is making, and my pass can
verify *how* those decisions are taxing the physical hardware and memory
sectors.
Looking forward to teaming up on this and keeping the foundation
bulletproof!
…On Sat, Jun 13, 2026 at 3:56 PM Olle Tiinus ***@***.***> wrote:
Hi Felix,
I’d be interested in helping, but in a fairly narrow area.
My strongest fit is provenance / replayability around agent and tool
execution: what decisions are made at runtime, what tools are allowed or
actually used, how model/provider choices are surfaced, and whether
user-facing behavior can be reviewed later without relying on chat history,
memory, or guesswork.
I’m not looking for broad maintainer responsibility. I think I’d be more
useful doing focused review passes on PRs, issues, or runtime paths where
agent behavior, tool execution, auth/secrets boundaries, model routing, or
recoverability matter.
I can start externally with a small pass over one concrete area and report
back with actionable findings: what seems observable, what seems missing,
what looks risky, and what I could or could not validate. I’ll be clear
about the difference between static review, local testing, and anything
that would need a different setup or hardware.
I’ve been doing public hands-on R&D around traceable AI workflows and
runtime observability, so this is a real area of interest for me, not just
a vague “happy to help.” But I’d rather start from Odysseus’ actual shape
and see whether the output is useful.
I work best with bounded requests: point me at a PR, issue, or runtime
path, and I’ll come back with findings rather than trying to take over a
broad area.
—
Reply to this email directly, view it on GitHub
<#551>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AQSPCWDHZZUR5BT42KIOLUL47VTTJAVCNFSNUABJKJSXA33TNF2G64TZHMYTENJVGE4DANRQGY5UI2LTMN2XG43JN5XDWMJQGE3TONBYG6QXMAQ>
.
Triage notifications, keep track of coding agent tasks and review pull
requests on the go with GitHub Mobile for iOS
<https://github.com/notifications/mobile/ios/AQSPCWAGADT2YNU5RBVJ6NT47VTTJA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZSHEYDEMRYUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSVGM33PORSXEX3JN5ZQ>
and Android
<https://github.com/notifications/mobile/android/AQSPCWG7Y4VH2V7JGRLZ73347VTTJA5CNFSNUABIM5UWIORPF5TWS5BNNB2WEL2ENFZWG5LTONUW63SDN5WW2ZLOOQXTCNZSHEYDEMRYUZZGKYLTN5XKOY3PNVWWK3TUUVSXMZLOOSXGM33PORSXEX3BNZSHE33JMQ>.
Download it today!
You are receiving this because you commented.Message ID:
<pewdiepie-archdaemon/odysseus/repo-discussions/551/comments/17290228@
github.com>
|
Beta Was this translation helpful? Give feedback.
-
|
Thanks, I think that split could make sense if a concrete PR or runtime path shows up. I’d prefer to keep it grounded and avoid forming a broad “team” shape before there is an actual target to review. The useful first step would probably be one bounded pass on a specific PR, issue, or agent/tool execution path, where we each mark what we can actually validate. My side would stay focused on provenance / replayability: what decisions are visible, what evidence survives, what is hard to reconstruct later, and where validation limits are. If your side can ground the lower-level execution or telemetry claims against the same concrete target, that could be complementary. |
Beta Was this translation helpful? Give feedback.
-
|
I am a hobbyist developer, and a retired Navy nuclear reactor operator with lots of time on his hands. I have very little experience, but all the time in the world to commit to learning and contributing to this. I think I have already developed some decent work practices by drawing on my engineering experience, despite those being two very different realms. I have forked the project and am working on staged contributions there before submitting issues and pull requests. I would love to become an active part of the community and to be brought inside the project if I am eventually seen as valuable enough for that. I am an Artix Linux user, which means s6 init instead of systemd. I believe that poises me to make sure things work across most Linux distros. I have set up Windows in a VM and am working on getting macOS working in a VM too, but none of that is integrated yet as part of my development workbench. I also have a tablet I can use to test remote access on low power devices and the touch screen interface. My goal is to become capable of testing most if not all use cases for my work properly before submitting it. Currently I think my biggest possible contributions are:
They are both complete and staged in my fork. I am doing a final sweep through them to make sure I am happy with my work before I submit them later today. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, I’m Zane Bartlett. I’ve been working as a software developer for a little over two years at a company building a pharmacy AI chatbot and related products. My role is focused on the layer between product data, model inputs, model outputs, and structured user-facing actions: shaping the context the AI receives, keeping responses parseable, and making sure the system behaves reliably enough for production use. I’m also very comfortable in Linux-heavy environments. Outside of work, I run Ubuntu as my main system and self-host over a dozen services with rootless Podman/systemd, reverse proxying, authentication, media workflows, backups, and other moving parts. That gives me practical experience with the setup, debugging, documentation, and reliability issues that tend to matter for a project like Odysseus. My personal site is https://zanebartlett.com, though it is a bit dated on my most recent projects. A more recent project is https://osmforbusiness.com; the About page has more context on what it is and why I built it. I don’t have formal open-source maintainer or reviewer experience yet. I do maintain the open-source website for the volunteer organization I’m involved with, but that is mostly solo work. I have made a few small suggestions and documentation fixes to larger repos such as VS Code, Airflow, and ngrok. Where I do have relevant experience is in volunteer and team coordination. I have over five years of volunteer management experience, worked for a nonprofit, and am currently on the board of one; so I understand the importance of clear communication, realistic commitments, follow-through, and being proactive when availability changes. Realistically, I could commit around 5 hours per week, and occasionally 10+ hours when needed or when I’m deep in a specific area, for very probably at least the next six months. I want to drive home that you can be confident in that estimate, because I know firsthand how much time and energy can go into onboarding volunteers, and how frustrating it is when people overcommit and disappear. I think I’d be a good fit because I am happy to help with the less glamorous but important parts of a fast-moving project: setup/docs improvements, issue triage, reproducing bugs, consolidating duplicates, keeping dependencies and instructions current, and helping make contribution/review workflows clearer. On top of my job giving me incentive to stay current, I’m also personally interested in getting deeper into AI systems technically, so this is an area I’m actively following and trying to grow in. I’m not claiming to be the strongest developer who will apply, but I do think I bring a strong mix of software experience, Linux/self-hosting experience, AI product experience, volunteer/team coordination experience, and steady operational reliability. |
Beta Was this translation helpful? Give feedback.
-
|
As i see open 688 open PRs i would right away block any AI generated slop, take notes from creator of Zig. Maintaining this kind of load is not possible and would probably bring more harm than good. |
Beta Was this translation helpful? Give feedback.
-
Intro, relevant experience15 years experience - 5 before getting a degree, 10 after. I started programming professionally, before going to school, working alongside a graphic designer, creating Wordpress admin backends (and some frontend widgets). Wordpress was 'new' back then. I obtained a BS in CS a few years into that narrow experience, and went directly into working for a startup. When I arrived, the entire team could fit on a couch. I was "the" developer and architect, working alongside great people serving as "the" big data guy, the artists, the issues-tracking and customer service staff, the CTO, the CEO, and the others who played their non-developer roles within the company. Now, the company has their name on a 10-story building + tertiary properties. I was responsible with coming to understand, maintain, triage, and translate into various microservice-based apps, an extensive in-house CRM/CMS - including SOAP and REST API calls (which if failing could mean major lawsuits), back and front ends of all interfaces across customers/clients, responsible parties, salespersons, executives, etc. I worked with SQL, PHP, Java, Javascript, Python, C++, Spark... Infrastructure as Code (ie. a lot of yaml and json has passed by these eyes). I've parsed tons of logs (eg. Cloudwatch), set them up to be formatted for big data analysis... I suppose with such a startup, Altogether, this system worked withdata from lab results, sales, promotions, HIPAA protected client data, life-critical monitoring services (RRMS), and more... I'll probably remember more things later, but that's the jist of the experience that most aligns. Over the first 2.5 years with the company, mostly as a contractor, I started with looking at a long list of existing bugs and feature requests. While numerous, requests were coming from a small team. By the end, I would become a part of a much, much larger ecosystem complete with scrum meetings, product owners, sprints... all that good stuff. Any prior open-source maintainer/reviewer experienceNot on Github no... We used private repos - svn, then AWS-hosted solutions. I did put in a lot of work to be proud of, working on bleeding edge CI/CD solutions for our growing, increasingly decoupled architecture. But our CTO would have been considered the maintainer/reviewer, ultimately, for most of my time in the relevant position. How much time you realistically haveerr.. hm... er... uh.. Why you think you’d be a good fitI have experience that aligns with the current state of the repo, and with a wide range of contributor experience levels - having trained many programmers, and worked alongside more experienced ones. Lots of experience working in teams, in physical and digital realms, I understand the pressures, lifecycles, and generally what to expect within this type of ecosystem. I was forced to learn and implement a lot of technologies, patterns, etc. very quickly... but I became a master of none (not to my own standard) - so I feel I wouldn't be too opinionated to be a fair maintainer. I'd be easy with the newcomers, understand the gravity of the experts, but also have enough self-run experience to understand issues - successes - failures; where they come from, and common ways to achieve them. I was always the go-to guy for hard-to-solve problems, as I have little issue with being tasked to 'almost overthink'. |
Beta Was this translation helpful? Give feedback.
-
|
Love what the whole premise of this Open Source LLM is. Here is a safeguard which blessed the GLM 5.2 benchmarks for long term alignment and mitigating hallucination. I will come back in a bit, just dealing with some IRL things. Please someone integrate this if possible. I see some Legends here, Love the vibes. Keep Cooking y'all. God Bless. Amen. |
Beta Was this translation helpful? Give feedback.
-
|
Follow on X, leave ur acc im on private. @KekiusWolf |
Beta Was this translation helpful? Give feedback.
-
This is all theory y'all know how it goes... Open source drop after certain things go down innit. Stay Blessed y'all.
|
Beta Was this translation helpful? Give feedback.
-
|
Hello, My name is Mathias. I currently work as a team manager and was a senior developer for 6 years for my current company (I manage as of this year, but I still code a lot as my team is still quite small). My expertise lies in a c# backend and angular frontend, so it is quite different than the current stack. I really love the approach with the local data and privacy that this project gives and I think it has a lot of potential, however I have only used it for about 2 days and I have seen a lot of bugs already. I am not sure where exactly they stem from but I would not be surprised if it is mostly frontend based with how it is currently implemented (I have never seen so wild vanilla js like this) 😆 . I would love to re-write it in something more modern like angular (what I know best, so it is my go-to). Using tailwind and primeNG on top or similar. I am planning on doing the re-write myself anyway, cause I love the idea of this project but can't stand the bugs, so either I can commit the angular frontend to the repo here and contribute directly or in my own forked version. Completely up to the owner/maintainers, just felt like sharing my thoughts here to see if there was any interest in collaboration! ❤️ I currently spent almost all my freetime on my training app, so building this would conflict a bit, but I am ready to share my time around 50/50 for the two projects for a start, so realistically around 10-15 hours a week or so. Timezone wise, I live in Denmark and will be moving to Japan from next of year if my current plans work out. Best Regards, |
Beta Was this translation helpful? Give feedback.
-
|
That is super cool! Awesome poc! Skimming through it a bit now! I am currently experimenting a bit with openwebui to see if it fits my use case. It seems they have added features very close to 'the brain' in Odysseus to Openwebui in the latest release, and is getting actively worked on too. |
Beta Was this translation helpful? Give feedback.
-
|
How closely is the frontend and backend connected? |
Beta Was this translation helpful? Give feedback.
-
|
Thanks! Not sure if I pushed the latest static version, but I adjusted the loading procedure -- and I currently don't land on '/chat' by default, so the app loads almost instantly. The legacy procedure, I think lasts 5 seconds minimum, per its timeout design. (I should keep up more with OpenWebUI's latest releases - the last one was impressive) I'm simultaneously working on a standalone node version too, just to practice / learn / gain experience with sveltekit. Of course it's not as far along, because of the many different ways the frontend calls the backend. Some places attempt to use an API_BASE prefix, many places just attempt to fetch "/api/..." with no prefix... and the legacy CORS .env option in the backend seems to not apply for the same reasons the frontend is difficult to update - many misaligned, non-standardized calls, missing usages of environment variables, etc... kind of a mess to work through. TLDR: The "adapater-static" option (svelte-dev) is a much easier route than "adapter-node" (svelte-node, if I've pushed it - probably not, until it's on par with svelte-dev). The backend changes: There should only be ...?two of changes to .py files for the svelte-dev poc. Mainly, just the change in app.py to serve the static svelte directory - and some usage of constants (eg. constants.py) to make it clean. Some of the other changes to .py files are general usability changes. For example, being able to choose my SearXNG port. There are also updates to backend tests for my refactor of the frontend ChatRenderer (making it only handle rendering chats) - see #4540 (PR). More info in #4047 - quick overview of what was changed. The former changes - svelte. The latter - for my own sanity. |
Beta Was this translation helpful? Give feedback.
-
|
Omg that frontend api stuff sounds like a pain to fix around the codebase 😩 That and of course many other things is what makes growing codebases so hard to maintain and continue developing on... Difference between short term and long term velocity... Sounds like you are making a lot of great updates to the whole app though I was also considering practising how to work with LLMs. Creating a c# backend that utilizes the semantic kernel that Microsoft has made. It sounds surprisingly good and it could really speed up the dev time for a homemade LLM app like this. Looking forward to see how your svelte kit project moves along. I can definitely recommend trying to look at openwebui. It is quite more extensive than I had imagined. |
Beta Was this translation helpful? Give feedback.
-
|
It is certainly painful. It would be so much easier if things were factored better (ie. separation of logic + concerns). I barely got to use c# professionally, so I built a whole suite of apps for a Unity project I'll leave unnamed - but I love that language now, so I understand. OpenWebUI has seen an insane number of commits over the last month. I'm a bit worried it's going to get bought out before long - same for the Continue vscode extension 😰 |
Beta Was this translation helpful? Give feedback.
-
|
Thanks everyone. I have sent collaborator invites with write access to: @CollinOS @alexkenley @undergroundrap @PovilasKirna @lalalune @RosenTomov @StressTestor @bitboody GitHub may show this as a pending repository invitation until you accept it. Appreciate the help. |
Beta Was this translation helpful? Give feedback.
-
|
Can you re-consider my application? I'd be excited to work on this. |
Beta Was this translation helpful? Give feedback.
-
|
@pewdiepie-archdaemon Could we please get feedback or a decision on this PR? https://github.com/pewdiepie-archdaemon/odysseus/pull/4101 I now actually have a lot more time also to contribute, since I am on parental leave for the next 40 weeks. |
Beta Was this translation helpful? Give feedback.
-
|
@Bandonker @Michiel-VandeVelde @NicholaiVogel @RaresKeY @Sigmanificient @alteixeira20 @mikaelaldy @nopoz @o3LL @vcscsvcscs @vdmkenny @CollinOS @alexkenley @undergroundrap @PovilasKirna @lalalune @RosenTomov @StressTestor @bitboody For maintainer coordination, please install Element Messenger. Sign up with this homeserver: Use registration code:
After signing up, send me your Matrix ID. It looks like:
I’ll manually approve/invite people into the Maintainers room. Please make sure your Element/Matrix name matches your GitHub username so I can verify who is who. |
Beta Was this translation helpful? Give feedback.
-
|
@undergroundrap:meet.dfx522.cc thank you! |
Beta Was this translation helpful? Give feedback.
-
|
@bitboody:meet.dfx522.cc |
Beta Was this translation helpful? Give feedback.
-
|
@PovilasKirna:meet.dfx522.cc |
Beta Was this translation helpful? Give feedback.
-
|
@nopoz:meet.dfx522.cc |
Beta Was this translation helpful? Give feedback.
-
|
@Bandonker:meet.dfx522.cc |
Beta Was this translation helpful? Give feedback.
-
|
Hey Pewds, I'm a developer with 10 years of experience at a small software company, with flexible hours and a lot of time for hobby projects. Odysseus has already become one of those projects for me. I have several open PRs that I’m actively maintaining, and I plan to keep contributing more. Odysseus is my first major public contribution, but I’ve been having a lot of fun with it, and I have ideas for how it could become an even better workspace, especially with tools that make it more useful for developers. I’d love to be part of the team helping with that. I also have some experience with AI and reinforcement learning systems, including video game bots using reinforcement learning in Unity, robotics control systems, and a custom workflow for AI coding agents which reduces hallucinations, context-window limitations, and conflicts between multiple collaborators or agents working on the same project. I’d be happy to share that and potentially help bring similar ideas into Odysseus. Thanks for your consideration. I’ll keep contributing either way!
|
Beta Was this translation helpful? Give feedback.
-
|
Heya! I’m Greg Procunier. I’m an Associate Principal Specialist Solution Architect at Red Hat in Canada, focused on OpenShift, OpenStack, Ansible Automation Platform, RHEL, hybrid cloud architecture, platform modernization, virtualization, automation, and operational security. I’d be interested in helping in a focused lane around Kubernetes / OpenShift deployment support and regression testing. In my day job I’m often demoing Red Hat OpenShift AI, especially vLLM-based model serving, and Odysseus looks like exactly the kind of straightforward, compelling self-hosted AI application that can consume an OpenAI-compatible model endpoint from a platform like OpenShift AI. The area where I think I can be most useful is not broad UI direction or trying to review every PR. It would be:
For prior open-source experience: I maintain public technical notes, demos, and lab projects around OpenShift, OpenStack, identity, automation, SELinux, and reproducible infrastructure patterns. I have not been an upstream maintainer of a project at this scale, so I would not overstate that. Where I can add value is careful review, practical deployment testing, issue triage in the platform/deployment area, and turning working patterns into clear docs. Time-wise, I can realistically contribute a few focused hours most weeks, with more time available in bursts when I’m already working through OpenShift AI / vLLM demo scenarios or testing deployment changes. Why I think I’m a good fit: Odysseus is already useful as a local/self-hosted AI workspace, but Kubernetes/OpenShift support could make it easier to run in enterprise and platform-engineering environments. I spend my time with customers who care about exactly the things that tend to break late in the process: security boundaries, persistent state, repeatable installs, networking, model endpoint integration, and day-2 operations. I can help keep that lane boring, documented, and testable. To be clear: I’d be contributing personally, not making an official Red Hat commitment. But my background lines up well with the Kubernetes/OpenShift deployment and regression-testing gap, and I’d be happy to help there if that would be useful. https://github.com/pewdiepie-archdaemon/odysseus/issues/5037 |
Beta Was this translation helpful? Give feedback.
-
|
Hey @pewdiepie-archdaemon — putting my name in for the security side. I know the first invites already went out, but the way I see it, I've been doing this job for a month already — just without the title. And it's been genuinely fun. 21 PRs merged so far, and a good chunk of them are security fixes:
Most of these came from reading the code hunting for authz gaps and path-handling issues, so the "security audits of auth, integrations, and tool execution" bullet in your post is exactly what I already do here for fun. Outside odysseus I have security-focused PRs in trufflehog and nuclei. Availability: 1-2 hours most days. I'm going to keep hunting bugs and reviewing security PRs here either way — without write access you can just watch me do it, with it I can actually help you keep the queue moving. |
Beta Was this translation helpful? Give feedback.
-
|
I can help by asking dumb questions about local security. I am convinced that security is a not just narrow expertise of chosen individuals, but a practice and common knowledge. For other things, the dot product of my ability and willingness is the following:
Unfortunately, my action points are limited to 2 hours per week at max if this is a voluntary help (and at min if it is not). I am interested in secure and easy UX for a local sandbox with multiple agents, tasks. projects, image/meme gens, connected to telegram and stuff, benchmarks for new local models (including my own). Something so easy and straihtforward children can play with, and without Here Be Dragons signs on every corner. My relevant experience is in industrial automation, I know how a feedback loop looks and can draw gradient descent on paper. OS is Linux for 10 years after about 10 on Windows (was top 3% StackOverflow for PHP for Windows in the past life). I care about money-life balance. Got more life and less money recently. Python - I was well-known troll there. AI - went to NN/AI conferences before Caffe got released - less math - more roleplay, scene and stuff. Obsessed with UX/DX - small usability fixes can clean up big discovery ways for those who are not used to survive in the woods. Any prior open-source maintainer/reviewer experience - 2606 open tabs, with majority of them being open source projects. https://github.com/abitrolly and https://gitlab.com/abitrolly - I lied - I don't have a life. :D How much time you realistically have - 2 hours per week for volunteering, and everything else if free for money making. Why you think you’d be a good fit - you absolutely need a useless person like me in the team, so that other members can feel that their imposter's syndrome is a BS. :) I am an invaluable asset in every team and Odysseus looks awesome so far with human touch behind it and that it doesn't have to beg for money like pi.dev other agent-based projects. I like communication in the open. There is a chance for me to become a better human and stop piss people off. Or at least it will be a honor to be banned by a famous YouTube blogger. :D But I also really hope we can get complicated security stuff with Linux namespaces more accessible to ELI5 crowd.
Challenge the changers. I hear you. ) Feature freeze? Proposal-driven development? I have no idea what is going to work well for AI driven development, but it is interesting to explore. |
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
-
Looking for maintainers / extra hands
First, huge thanks to everyone opening PRs,
issues, suggestions, testing notes, and bug
reports. The amount of help around Odysseus
has been genuinely awesome.
I’m working through as much as I can, but
the repo is moving fast and there are now a
lot of PRs/issues to review properly. I’d
like to bring in a few experienced people
who can help keep things moving without
lowering the bar.
Areas where help would be especially
useful:
regressions
Docker, and GPU setups
duplicates
integrations, agents, and tool execution
If you’re interested, please reply with a
short intro:
experience
I’m especially looking for people who can
move quickly, communicate clearly, and be
careful with user-facing changes. The goal
is not just to merge more, but to keep the
project stable while it grows.
Thanks again. This has been a lot, but in a
good way.
Beta Was this translation helpful? Give feedback.
All reactions