AI / ML builder turning rough ideas into working product demos.
I build AI products that people can open, test, and understand: agent simulations, RAG systems, natural-language data tools, and polished portfolio prototypes.
Portfolio | Repositories | Latest AI DevOps prototype
I focus on small but complete AI systems: clear UI, realistic demo data, safe execution boundaries, and enough backend logic to prove the idea. My portfolio work is built around practical workflows instead of static mockups.
- Agent-based simulations for marketing and DevOps decisions.
- RAG and retrieval comparison systems with measurable outputs.
- Natural-language data interfaces with SQL validation and source-row inspection.
- Browser-first demos that can be tested without private keys.
AI DevOps control-plane simulation for testing deployment flows. It runs Harness-style CI/CD scenarios, shows pipeline logs, generates AI-style analysis, and keeps tester run history.
|
Multi-agent campaign simulator for validating marketing ideas before launch. It creates persona-based user agents, runs campaign behavior, and returns a visual performance report.
|
Full-stack RAG strategy comparison prototype. It compares keyword, vector, hybrid, memory, and graph-style retrieval on the same datasets with evidence, latency, coverage, and scoring.
|
Natural-language company-data query demo with deterministic intent compilation, safe read-only SQL, visible source rows, and an experimental schema-aware SLM path.
|
I am shaping a portfolio around AI products that can be evaluated hands-on: not only prompts and screenshots, but systems with workflows, state, validation, and deployment. The goal is to make every project understandable in the first minute and testable in the next five.
- DeployPilot - AI DevOps simulation UI for deployment flow testing.
- AdPilot - Multi-agent campaign simulator.
- RAGLab - RAG strategy comparison lab.
- DataTalk - Natural-language company-data query demo.