I'm working with an early-stage, stealth London startup building a real-time AI copilot for gamers. Voice + vision + a deep game knowledge engine, so the kind of product that should already exist and somehow doesn't.

There are three layers stitched together: a knowledge engine that understands games deeply (mechanics, items, builds, boss strategies, hidden secrets), real-time visual understanding of what's on screen, and a natural voice interface that ties it all together.

They're hiring across three specialisms - you need depth in one:

• Voice: STT–TTS, WebRTC, real-time streaming
• Vision: multimodal models, embeddings, entity extraction
• LLM / RAG: retrieval, knowledge graphs, agentic orchestration

Plus: strong TypeScript / Node / Postgres, architectural instincts, coding-agent native, and you play a lot of games in your spare time.

Details

• London-based, existing UK right to work ...