I will audit and optimize your rag vector search performance

V
valhallasoft
V
valhallasoft
Martin Poli
audit and optimize your rag vector search performance
Full Screen
audit and optimize your rag vector search performance
Full Screen
Full Screen
audit and optimize your rag vector search performanceaudit and optimize your rag vector search performanceaudit and optimize your rag vector search performance

About this gig

Your RAG is in production but returning bad results. Latency is slow. Costs are climbing. Hallucinations slip through. Sound familiar?


I audit and fix RAG pipelines that look right on paper but fail in the real world. 10+ years of production backend work, currently leading the AI search migration for one of Latin America's largest retailers (50K+ products, 1M+ daily users).


What I audit:

  • Embedding model fit for your domain
  • - Chunking strategy and overlap
  • - Retrieval recall and precision (with eval set)
  • - Reranking effectiveness
  • - Hybrid search weights (keyword vs semantic)
  • - Latency per stage and cost per query
  • - Hallucination patterns

What you get:

  • Written diagnostic with prioritized fixes
  • - Code changes for top issues (Standard / Premium)
  • - Eval set so you can measure progress
  • - Monitoring setup (Premium)

Stack: Python, OpenAI, Anthropic, Pinecone, Weaviate, Qdrant, pgvector, LangChain.


Send me your stack and one example query that fails. I will tell you what is likely broken before you pay.

  • AI engine

    • GPT
    • Gemini
    • Bert

  • Programming language

    • Go
    • JavaScript
    • PHP
    • Python
    • TypeScript

Get to know Martin Poli

Martin Poli

Senior RAG and AI Search Engineer for Backend at Scale

  • FromUruguay
  • Member sinceMar 2020

  • Languages

    English
Senior Platform Engineer with 10+ years building production systems at scale. Currently leading platform infra and AI search for Argentina's largest retail chain (200+ stores, 1M+ users/day), replacing Google Search API with RAG-based semantic search across 50K+ products. What I do best: - RAG, embeddings, OpenAI/Anthropic/Bedrock - Vector DBs: Pinecone, Weaviate, Qdrant, pgvector - Backend at scale: Python, Go, Node.js, PHP 8 - AWS EKS, Karpenter, Terraform, multi-account IaC Have a search problem or an LLM pipeline that won't ship? Send me your stack.

My Portfolio