I build applied AI systems and full-stack products with a focus on grounded retrieval, reliable automation, and practical user workflows.
Right now, I am exploring multilingual RAG, document intelligence, AI-assisted education, and local-first agent tooling. I care about evaluation, traceable outputs, and turning prototypes into software people can actually use.
| Project | What it demonstrates |
|---|---|
| Document Assistant | An evidence-first multilingual RAG workspace with hybrid retrieval, citations, evaluation, and document intelligence tools. |
| QuestLine | A Persian-first coaching operations platform built with Next.js, NestJS, Prisma, and PostgreSQL. |
| AI Exam Platform | A class-centric assessment system with teacher-controlled AI grading, review, and result publication. |
| Resume Intake Agent | A LangGraph and Ollama workflow for resume extraction, screening, interview scheduling, and approval-gated communication. |
| Mission Control OS | A local-first dashboard plugin for Hermes Agent with runtime visibility, approvals, conversation tools, and voice hooks. |
- Ground AI outputs in inspectable evidence.
- Prefer small, testable changes over opaque demos.
- Keep human review in consequential automation loops.
- Design Persian and RTL experiences as first-class product surfaces.
Python · TypeScript · FastAPI · Next.js · NestJS · PostgreSQL · Qdrant · LangGraph · Docker
If you are working on document intelligence, applied AI, or developer tooling, feel free to open a discussion or issue in the relevant repository.
