AI Engineer

We are hiring a remote AI Engineer with 2 or more years of software experience to build LLM features, agent workflows and the evaluations that show they work, in client projects and our own products. ## What you will do - Build LLM features into products: drafting, summarising, pulling structured data out of messy text, and assistants that call tools. - Build agent workflows with clear limits: which tools they can use, what they may spend, and human review before anything public or irreversible. - Write evaluations: test sets built from real cases, regression checks before a prompt or model change ships, and numbers for accuracy, speed and cost. - Pick the right model for each job, route requests across providers with fallbacks, and keep costs predictable. - Use retrieval (RAG) where it helps, and say so when plain search or a simple rule would do better. - Fine-tune and run small open models when privacy or speed means the data should stay on the device. ## What we look for - 2 or more years as a software engineer, in Python or TypeScript (ideally both). - At least one LLM feature you shipped to real users, and an answer to “how did you know it worked?” - Evaluations as a habit: you measure before and after a change. - A clear view of the failure modes, such as made-up answers, prompt injection and data leaks, and designs that work around them. - Care with personal data: it goes only where it should. - Clear written English. ## Good to have - Fine-tuning (for example LoRA) - On-device inference (llama.cpp, GGUF, MLX) - Model Context Protocol (MCP) - PostgreSQL with pgvector ## How we work - **Written scope first.** Every piece of work starts with a short written scope: what ships and when. - **Weekly demos.** We show progress on a live link every week, to each other and to clients. - **A direct line.** You talk to clients and to our founder directly. There is no account layer in between. - **Remote and written.** We work remotely and write things down: decisions, handovers and runbooks. Expect a few hours a week of overlap with India time (UTC+5:30) for demos and reviews. ## How to apply Email careers@roundrobin.in with "AI Engineer" in the subject line, and send: - A link to an LLM project you built (repository, demo or write-up), and how you evaluated it. - A short note on a failure you found in an LLM system and how you fixed it. - Your CV or LinkedIn profile, your time zone, and when you could start. - **We read and reply.** We read every application and reply to each one, including when the answer is no. - **A conversation.** If there is a fit, a call of about 30 minutes about work you have done. - **A clear answer.** We decide and tell you either way. We never charge applicants anything. Apply by email: careers@roundrobin.in Role page: https://roundrobin.in/careers/ai-engineer/