Abstrabit is a private AI company based in Bengaluru. We help businesses run AI entirely on their own infrastructure: self-hosted LLMs, secure enterprise RAG, fine-tuned models, and AI agents that never send data outside the network. We serve clients across the US, UK, and EU, and our work spans the full private AI stack, from GPU infrastructure to production agents.
About the Role
We are hiring interns who are strong in any one of these four areas. You do not need all four. Pick your strongest layer and apply.
LLMOps: serving open-weight models with vLLM or SGLang, GPU infrastructure, quantization, Kubernetes deployment, latency and cost benchmarking
Enterprise RAG: ingestion and chunking pipelines, embeddings, hybrid retrieval and reranking, retrieval quality evals
Model fine-tuning: dataset curation, LoRA/QLoRA and full fine-tunes, instruction tuning, eval design
AI agents & workflows: multi-step agents, tool use, orchestration, memory, guardrails, workflow automation
Requirements
What You'll Do
Build and ship working systems, not slide decks
Own your work end to end: design, build, benchmark, document
Work directly with the founders and senior engineers
Go deep in your layer while getting exposure to the other three
What We're Looking For
Proof of good work in your chosen area. That could be:
A project you built and can demo
Open source contributions
Kaggle, GSoC, hackathons, or a solid technical write-up
Anything public and verifiable works. College and CGPA don't matter. What you've built does.
Beyond that: you read documentation, pick up new tools fast, and figure things out without waiting to be told.