hanzala122/rf-telecom-rag
0
RF / Telecom RAG
Chat with 39,818 arXiv eess.SP abstracts. Ask about OFDM, MIMO, RIS, beamforming, NOMA, 5G/6G, radar, channel estimation, signal processing.
Stack
- Retrieval: bge-small-en-v1.5 (384-dim) + Neon PostgreSQL with pgvector (HNSW index)
- LLM: Llama-3.1-8B via Groq
- Tracing: Langfuse Cloud
- API + UI: FastAPI + vanilla HTML/CSS/JS chat (no React build needed)
Endpoints
/- Chat UI (NASA mission-control aesthetic)/query- Non-streaming JSON RAG endpoint/query/stream- Server-Sent Events streaming endpoint/health- Liveness probe/docs- Auto-generated OpenAPI / Swagger UI
Notes
- First request after idle may take ~20s (Neon free-tier compute auto-pauses)
- Built as Week 2 of a 6-week LLM engineer sprint by @hanzala122
