Debugging & Optimizing RAG Pipeline for Radiology Chatbot
Budget: $2 – $8 USD
Debugging & Optimizing RAG Pipeline for Radiology Chatbot
We have a production RAG (Retrieval-Augmented Generation) chatbot for radiology built with Python/FastAPI backend, Next.js/React frontend, Qdrant vector DB, and multiple LLMs (Claude, Gemini, Groq).
Key tasks:
1. Optimize post-synthesis citation verification pipeline (currently 3-10 min, need under 15 seconds) - batch API calls, fix Semantic Scholar auth, remove redundant passes, cap gap-fill retries
2. Fix Docker deployment pipeline - code changes not reaching running container, need proper volume mounting or rebuild workflow
3. Improve citation quality - LLM produces great citations but verification strips them; need LLM-first verify-second approach
4. Reduce pipeline complexity - consolidate overlapping modules, remove dead code paths
5. Establish config management - sync local and VPS environment variables
Tech stack: Python, FastAPI, Next.js, React, Qdrant, Docker, Anthropic Claude, Google Gemini, Exa.ai, Tavily, PubMed, Semantic Scholar, Cohere
Looking for someone experienced with RAG pipelines, Python backend optimization, and Docker deployments.
We have a production RAG (Retrieval-Augmented Generation) chatbot for radiology built with Python/FastAPI backend, Next.js/React frontend, Qdrant vector DB, and multiple LLMs (Claude, Gemini, Groq).
Key tasks:
1. Optimize post-synthesis citation verification pipeline (currently 3-10 min, need under 15 seconds) - batch API calls, fix Semantic Scholar auth, remove redundant passes, cap gap-fill retries
2. Fix Docker deployment pipeline - code changes not reaching running container, need proper volume mounting or rebuild workflow
3. Improve citation quality - LLM produces great citations but verification strips them; need LLM-first verify-second approach
4. Reduce pipeline complexity - consolidate overlapping modules, remove dead code paths
5. Establish config management - sync local and VPS environment variables
Tech stack: Python, FastAPI, Next.js, React, Qdrant, Docker, Anthropic Claude, Google Gemini, Exa.ai, Tavily, PubMed, Semantic Scholar, Cohere
Looking for someone experienced with RAG pipelines, Python backend optimization, and Docker deployments.