LLM Integration Expert to integrate Claude/GPT-4 into our SaaS Platform (RAG, Next.js, Python)
Budget: $150.0
FIXED /
⭐ 5.00 (149)
United States
node.js, python, api-integration, react-js, typescript
Our SaaS platform has served enterprise clients for years and recently shipped its first AI
assistant. That was step one. Now we are going deeper: we want every text heavy workflow in
the product to be AI assisted, and we need one solid foundation to build it all on.
The mission for this contract: build a provider agnostic LLM layer inside our platform and
connect it to real product workflows. Think document summarization, semantic search over
customer data with RAG, smart auto fill, and AI generated reports. One clean integration layer,
many features hanging off it. Claude first, GPT-4 as fallback, and the freedom to swap providers
without rewriting features.
The stack is Next.js/React/TypeScript on the front, Node.js and Python services on the back,
PostgreSQL, multi tenant, cloud hosted, real users at real scale.
What matters to us:
You have done this on a live production system, not a weekend demo. You have dealt with the
unglamorous parts: tenant data isolation, token cost budgets, provider outages and rate limits,
structured outputs that actually stay structured, and AI that fails gracefully instead of taking a
feature down.
You should be comfortable with:
● Claude API and OpenAI API, including tool/function calling and structured outputs
● RAG pipelines: chunking strategies, embeddings, vector search (pgvector, Pinecone or
similar)
● Node.js and Python service development inside an existing codebase
● Streaming responses end to end, from the API to a Next.js UI
● Multi tenant isolation, caching, and per tenant token cost tracking
It would be better if you also know LangChain or LlamaIndex, Vercel AI SDK, AWS Bedrock or
Vertex AI, and have run LLM evals (LangSmith, Braintrust or similar).
How to apply:
Skip the generic pitch. Start your proposal with the word "Orbit" so I know you actually read this,
then tell me in a few lines about the deepest LLM integration you have personally shipped and
what made it hard. Links to live work beat certificates.
Ouvrir sur Upwork