About PipesHub

PipesHub is the open-source Context Layer for Enterprise AI. Connect enterprise knowledge across your organization, preserve access permissions, generate trustworthy citations, and build AI agents, enterprise search, RAG applications, MCP servers, and agentic workflows on a single governed context layer.

Our AI stack is Python and FastAPI, LangChain and LangGraph for pipelines and agent workflows, Qdrant for vector search, Neo4j or ArangoDB for the knowledge graph, Kafka and Redis Streams for events, Celery for background jobs, Docling and PyMuPDF for parsing documents. We're model-agnostic, so any LLM provider, running inside the customer's own VPC.

You'll ship code into a public repo that other engineers read, fork, and file issues against.

What you'll work on

We'll shape this around what you're valuable at, but expect a mix of:

- Retrieval quality. Chunking, embedding choices, hybrid search, reranking, and stitching vector search together with graph traversal.
- Agents. Building and improving LangGraph workflows: tool calling, structured outputs, multi-step planning, recovering when a step fails.
- Document understanding. Turning PDFs (including scanned ones), spreadsheets, slides, images, audio, and video into clean citable blocks.
- Evaluation. Building the eval sets and tracing that tell us whether a change actually improved answers, instead of guessing.
- Grounding. Making sure every claim maps back to a real block in a real document, and catching it when it doesn't.
- MCP and SDKs. Extending our MCP server and the Python, TypeScript,



and Go SDKs.

What we're looking for

- Strong Python. You've written and debugged real code, not just followed a course.
- You've built something with LLMs beyond chatting with one. An API integration, a RAG pipeline, an agent, a fine-tune. Something you can walk us through.
- A working understanding of embeddings, vector search, and prompting.
- Familiarity with LangChain, LangGraph, LlamaIndex, or something comparable.
- Git, and enough confidence to read an unfamiliar codebase without hand-holding.
- Final-year students, recent grads, and self-taught builders are all welcome.

Nice to have

- A public GitHub or a deployed demo. We care about this far more than the length of your CV.
- Open-source contributions. A PipesHub PR is the strongest possible signal.
- FastAPI, Docker, Qdrant, Pinecone, Weaviate, Neo4j.
- Any exposure to knowledge graphs, information retrieval, or search ranking.
- Eval and tracing tools like LangSmith, Ragas, or Phoenix. Your own harness counts too.

What you get out of it

6 months on production AI systems at enterprise scale. Permission-aware retrieval across 30+ connectors is a harder problem than most people touch this early in their career.

You get mentorship from the engineers who built the platform, public commits you can point at for years, and a real shot at a full-time offer.

Apply: https://forms.gle/UH514zy89P46TS3p9

Include links to what you've built. One well-explained project beats 10 half-finished ones.

Stipend: 50k - 80K INR

📌 AI Engineer Intern (Bengaluru)
🏢 PipesHub
📍 Bengaluru