Browse categories
Explore
Fiverr Pro
English
$
USD
Looks Like This Service Is On Hold
I will build a rag ai search pipeline with elasticsearch
Serhii T
About this gig
Want to add AI-powered search or a Q&A bot to your product or internal data? I build production-grade RAG pipelines that actually work not toy demos.
What I offer:
- RAG pipeline: ingest data chunk embed index retrieve answer with citations
- Hybrid search: BM25 + vector with Reciprocal Rank Fusion
- Intelligent query routing: simple queries = fast path ($0 LLM cost), complex = AI
- LLM integration via LiteLLM: Claude, GPT, DeepSeek with fallback
- Agentic workflows with LangGraph: multi-step reasoning, tool use
- Full observability: Langfuse tracing, cost tracking per query
Why me:
- 8+ years Elasticsearch in production retrieval is where most RAG projects fail
- Built production RAG Q&A systems, not hackathon demos
- Designed AI orchestration engine with triage routing (Fast/ReAct/Planning)
Data sources: PDF, DOCX, databases, APIs, CRM data anything text-based.
Stack: Elasticsearch, Qdrant, LangGraph, NestJS, Vue.js, LiteLLM, Docker.
Message me with your use case I'll tell you if RAG is right or if simpler search will do.
Get to know Serhii T
Serhii T
Elasticsearch, AI Search Engineer, 8 Years Production Experience
- FromPoland
- Member sinceApr 2026
- Last delivery3 months
Languages
English, Ukrainian, Russian
Elasticsearch & AI Search Engineer, 8+ years production experience. I deploy, optimize, and scale search infrastructure for enterprise systems.
Core skills:
- ES deployment, performance tuning, relevance optimization
- Multi-tenant search architecture
- RAG pipelines & AI search (LangGraph, Qdrant)
- Docker & Kubernetes for search clusters
Stack: Elasticsearch, TypeScript, NestJS, Java, Docker, Docker-Compose, K8s, Helm, RabbitMQ, Vue.js, LangGraph.
Based in Poland. Open to consulting and projects.
My Portfolio
FAQ
What data sources can you work with?
Documents (PDF, DOCX), databases (PostgreSQL, MySQL), APIs, CRM data, web content — anything text-based.
How do you keep LLM costs low?
Intelligent triage routing. 60-70% of queries are answered by fast hybrid search with zero LLM calls. Only complex questions use AI.
Can you integrate this into my existing app?
Yes. I deliver via REST API, TypeScript SDK, or embeddable web component — whatever fits your stack.
What if I'm not sure RAG is the right solution?
Message me first. I'll assess your use case for free and tell you honestly if RAG makes sense or if simpler search will work better.
