Senior AI Engineer

PokerDen is hiring an AI engineer to design, build and ship the game's first AI features, and to run them reliably in production. PokerDen is a social Texas Hold'em game: players create an account, take a seat and play with friends in real time. It uses play chips only, with no real-money betting. The working MVP has a full Hold'em engine, real-time tables over Socket.io, and player accounts with chip balances saved in MongoDB. It has no AI features yet. You'll be the person who decides how AI fits into the game and gets it into players' hands - What you’ll do You'll own PokerDen's AI work from prototype to production. In your first months, that means: 1. Build AI opponents. Create poker bots at several skill levels so players always find a game, even at quiet tables. Bots are always clearly labelled as bots, and they see only what a human player in their seat would see. 2. Protect fair play. Detect collusion, chip dumping, multi-accounting and players using bots or outside help, using game history and behaviour signals. Flag cases for review instead of banning automatically. 3. Moderate chat. Filter abuse, spam and personal information in lobby and table chat, with low enough delay that conversation still feels live. 4. Add a hand review assistant. After a hand ends, players can ask an LLM to explain what happened and what they might have done differently. It never gives advice during a live hand. 5. Make it production-ready. Each feature ships with evaluation sets, monitoring, cost and latency budgets, fallbacks for when a model or provider fails, and a way to roll changes back You'll work with the full-stack engineer, who owns the game server, and directly with the project owner on what to build next. What you bring - 4+ years in software engineering, with 2+ years shipping machine learning or LLM features that real users rely on - Strong Python, and enough TypeScript or JavaScript to work on our Node.js game server - Hands-on experience building with LLM APIs: prompt design, tool use, structured outputs, and evaluating model quality - Experience taking AI features to production: serving, monitoring, cost control, and handling model failures safely - Solid machine learning basics: classification, anomaly detection, and choosing the right metrics for imbalanced data - A habit of measuring: you build test sets before you tune, and you can show whether a change made things better - Good judgement about player trust and privacy: what data to use, what to log, and when a person should make the final call - Clear written English for async work