Browse categories
Explore
Fiverr Pro
English
$
USD
I will build a rag chatbot that answers from your documents
Leon
About this gig
I build production-grade RAG chatbots that turn your PDFs, docs, and websites into a knowledge agent your users can actually trust.
What you get:
- LangChain pipeline + Chroma or Pinecone vector database
- - OpenAI or Anthropic Claude embeddings + retrieval
- - FastAPI /ask endpoint or Streamlit UI
- - pytest evaluation harness with golden questions (no demos, real tests)
- - Langfuse observability so you can debug every retrieval
This is not a 2-hour ChatGPT wrapper. Every delivery includes:
- Source-cited answers (citations linked to source chunks)
- - Tested chunking + embeddings pipeline
- - Clean code you can hand to your team
- - 30-day bug-fix support on Premium
Stack: LangChain, OpenAI, Claude, Pinecone, Chroma, FastAPI, Streamlit, Langfuse, Railway.
Use cases: customer support FAQ bot, internal knowledge agent, document Q&A, contract or legal search, research summarizer, technical docs chatbot.
Tell me your corpus type (PDF / web / markdown), update frequency, and deployment target. I will recommend the right tier and ship in 4 to 14 days.
Get to know Leon
Leon
AI Systems Architect building production AI agents with LangGraph and RAG
- FromChina
- Member sinceJun 2024
Languages
English, Chinese
AI Systems Architect building production-grade AI agents for product teams. I design LangGraph and CrewAI multi-agent systems with RAG, tool use, and evaluation pipelines, not demos. Every delivery includes tested code, observability with Langfuse traces, and post-delivery support so you can debug and iterate. Specialties: knowledge agents, workflow automation, data extraction, vertical-domain AI. Fluent in English and Mandarin. Let's ship your AI system.
My Portfolio
Other AI Development Services I Offer
FAQ
Do you use AI tools in development?
Yes, Claude Code-augmented + human review on all deliverables. This is why delivery is fast and code quality is consistent.
What LLM provider do you support?
Claude (Anthropic), OpenAI, and local models via Ollama. I recommend Claude for highest accuracy on structured Q&A; OpenAI for ecosystem compatibility.
Will you include an evaluation or test harness?
Yes. Basic includes 10 golden questions, Standard 30, Premium 50. All pytest-based with pass/fail report so you can verify retrieval quality before going live.
How do you handle my data and API keys?
Client provides keys in .env (never committed). Source documents are deleted from my workstation after delivery plus a 7-day support window.
Can I update the corpus after delivery?
Standard and Premium include re-ingest scripts (auto re-ingest pipeline on Premium). Basic is one-shot ingest; corpus updates are available later as a custom offer.
