RAG Debugging Assistant
Paste an error from a backend, machine learning or data project. The assistant finds the closest known errors in its knowledge base, then an LLM writes a short fix using only those records. The matched records are shown next to the answer.
prashant@linux:~$ rag.retrieve(query)
Ask something to see the retrieved sources.
Knowledge Base
500+ generated error logs spanning backend APIs, ML, and data engineering, paired with exact technical fixes.
Architecture
TF-IDF retrieval with cosine similarity, paired with llama3.2 via Groq API for precise answer generation.
Pipeline
Vectorizes user queries to retrieve the top 3 matching logs, forcing the LLM to synthesize facts without hallucinating.