ProjectIntent Over Keywords: From Stitched Embeddings to a Production Fashion Agent · Module 16: Multimodal product search and ingestion · part 18 of 26
Part 18 · Module 16: Multimodal product search and ingestion

Module 16: Multimodal product search and ingestion

24 min read·
About 50 minutes · Needs: Module 7 (one embedding space), Module 13 (catalogue loader), Module 14 (the planner) · Code: part3-styleme/backend/src/styleme/agent/implementer.py, ingest.py, llm.py, vectorstore.py, scripts/audit_catalog.py

Where this fits

WHERE THIS FITS Intent Over Keywords this module PART 1: STITCHED VIDEO RAG three models, two spaces, RRF PART 2: UNIFIED VIDEO RAG one model, one space PART 3: STYLEME FastAPI, LangGraph, React QUALITY AND OPERATIONS evals, Docker, Cloud Run Frames ffmpeg, one per 5 s CLIP ViT-B/32 frames + text, 512 d Whisper base speech to text MiniLM 64-word chunks, 384 d Two collections p1_visual, p1_text Rank fusion RRF, k = 60 Gemini answer from transcript text API + Streamlit upload, ask, sources Windows ffmpeg, video + audio Embedding 2 Gemini, one vector One collection p2_unified, no fusion Notes + answer Gemini, cited times React web chat, photo, chips nginx one URL, /api proxy FastAPI /api/chat, limits Sessions in memory, cap of 3 Planner intent before retrieval Implementer batched embed, search Generator stylist reply, ids Grounding check no LLM, drops invented Gemini flash-lite, embedding 2 Qdrant styleme_products Catalogue ingest image + title vectors Langfuse optional traces Test video 26 questions, 4 kinds Video eval Part 1 vs Part 2 Agent evals intent, grounding Guards uploads, limits, secrets Containers compose profiles Cloud Run StyleMe only CI/CD GitHub Actions, WIF Original code v1, v2, fashion-agent
This module (highlighted): the implementer, the catalogue ingest and the Qdrant collection they share. Blue is our code, amber a model.

This module is the retrieval half of StyleMe. The planner (Module 14) has decided that a turn is a search and has written up to four queries. Here those queries, and the shopper's photo if there is one, become vectors, the vectors meet the catalogue in Qdrant, and a short list of products comes back for the generator (Module 17). We also cover how the catalogue got into Qdrant in the first place, and how one bad row was found and removed.

The rest of this project is yours to keep

This project is bought on its own, once, and stays readable afterwards, including the parts added to it later.

Buy this project · ₹799See what is inside

Or open everything with a plan