AI Engineer | NLP, RAG Systems, and LLM Applications
I help teams and clients turn documents and research ideas into production-ready AI systems — especially multi-turn RAG assistants, hybrid retrieval pipelines, and deployable LLM backends.
🟢 Available for freelance / contract work
📧 amir4javar@gmail.com
- Custom RAG chat assistants over PDFs / internal docs
- Hybrid retrieval systems (vector + keyword search)
- Multi-turn conversational AI with session memory
- LLM backends with FastAPI, WebSocket streaming, and Docker deployment
- Model efficiency work (distillation / optimization) when performance or cost matters
If you need an AI feature that is more than a notebook demo, I can design, implement, and ship it.
Live demo: Hugging Face Spaces
A deployable document Q&A system, not a toy chatbot:
- Multi-turn conversations with context-aware follow-ups
- Intelligent query routing (skip unnecessary retrieval)
- Hybrid search via Weaviate (dense vectors + BM25)
- Real-time token streaming over WebSockets
- FastAPI backend + Streamlit UI
- Auth / persistent sessions (Supabase)
- Docker deployment, automated tests, evaluation harness, optional tracing
Stack: LangGraph · Weaviate · FastAPI · Streamlit · Docker · OpenAI-compatible LLMs
Client use cases: internal knowledge bots, technical document assistants, policy/manual Q&A, research PDF copilots
- Retrieval-Augmented Generation — Production multi-turn RAG with hybrid search, LangGraph routing, streaming responses, and deployment path. [live demo]
- Born-Again Networks & Self-Distillation — Compared BAN / layer-wise self-distillation on CIFAR-100; +1.4% accuracy over baseline through iterative self-teaching.
- Multi-Teacher Distillation — Studied soft-label aggregation on GLUE; published a negative result showing teacher diversity matters more than teacher count.
| Area | Tools & Frameworks |
|---|---|
| ML / NLP | PyTorch · Hugging Face Transformers · Fine-tuning · Knowledge Distillation · LLM APIs |
| RAG & Agents | LangChain · LangGraph · Weaviate · Hybrid Search (vector + BM25) |
| Backend & MLOps | FastAPI · Docker · Linux · REST / WebSocket APIs · vLLM / Ollama |
| Frontend | Streamlit |
| Quality | Pytest · Evaluation (RAGAS / DeepEval) · Observability (OpenTelemetry / Phoenix) |
I take freelance projects in:
- RAG / document assistants (MVP → production)
- LLM application development (API + UI + deployment)
- Retrieval quality improvements (chunking, hybrid search, routing, evals)
- Efficient ML (distillation / optimization) when relevant
Typical engagement styles:
- fixed-scope MVP
- production hardening of an existing prototype
- short contract to ship a specific AI feature
If you have a project that needs an AI engineer who can ship:
- 📧 Email: amir4javar@gmail.com
- 💼 LinkedIn: linkedin.com/in/amirhossein-javartani-6622982b3
- 💻 GitHub: github.com/amir4javar
Open to freelance, contract, and strong full-time opportunities in applied AI / NLP.