About the role
You will build AI into real products: for clients adding AI to what they already run, and for our own platform, which uses it for question writing, grading suggestions and class notes. That means choosing the right model, writing and testing prompts, building retrieval and transcription pipelines, and the careful software that makes them safe to ship.
Responsibilities
- Build features on large language models such as OpenAI, Claude and Gemini, from the first prototype to production.
- Write prompts and structured outputs, and test them on real examples before anything ships.
- Build retrieval, transcription and document pipelines in Python or TypeScript.
- Measure quality, cost and speed, and keep each within what the client agreed to.
- Tell clients plainly what AI can and cannot do for them.
Requirements
- 2+ years in software engineering, with at least one AI feature you shipped to production.
- Python, and working TypeScript.
- Hands-on experience with LLM APIs, embeddings and vector search.
- The habits that make AI safe to ship: tests, evaluation sets, version control and reading logs.
- Clear written and spoken English.
Nice to have
- PyTorch or TensorFlow, speech-to-text, or fine-tuning open models.