wrap respond in try/except for error visibility

fix: handle None history in Gradio 5; catch all RequestException

build index from resume_text.txt at startup; remove reranker; split FAISS/BM25 queries

remove reranker, split FAISS/BM25 queries, 500-char chunks

improve prompts: strict career agent, expanded entity keywords

remove sources, add RRF, query rewriting, conversation history

switch to router endpoint with HF_TOKEN check

chore: remove unused imports

fix: restore missing retrieval functions

revert to api-inference endpoint (works on HF infra), switch to Qwen2.5-0.5B

fix: single chat function with proper API signature

improve prompt for metric-rich responses, remove :fastest suffix

rebuild index with latest NIQ CV only

fix: change model to Llama-3.2-1B-Instruct for router endpoint compatibility

fix: migrate api-inference to router endpoint

Upload 3 files

Update app.py

Update app.py

Rename faiss_store_new.zip to faiss_store.zip

Upload 4 files

initial commit