AI/ML Engineer - Live Games

About the role

Supercell's AI team is looking for an AI/ML engineer who turns modern models into features and tools that work in production at Supercell scale, under real deadlines. The role focuses on the gap between a model that works in a notebook and a feature trusted by game teams and hundreds of millions of players. You will build and ship AI features in live games and internal tools, working closely with AI Platform Engineers, AI/ML engineers, Clients Programmers and Cloud Engineers, and with Data, QA, Engine and game teams.

Key responsibilities

• Build and ship AI features and systems end to end into live games and internal tools, at production quality and under real deadlines.
• Own evaluation as a first-class deliverable: write and maintain benchmarks, evals and regression suites.
• Fine-tune, adapt and distill models where it pays off, and judge when a better prompt, retrieval or smaller model beats training.
• Make trade-offs on inference, including on-device, against parameter, latency and battery targets.
• Tackle hard applied problems such as making autonomous agents reliable and accurate enough to trust with real work.
• Work close to game and central teams, sometimes embedded with them, and build things that stay in use after you leave.
• Mentor junior AI/ML engineers through code review, pairing and design discussion.
• Stay at the frontier by reading the literature and major conferences and bringing back what is genuinely useful.

Required profile

• Strong, current command of modern generative methods at the applied level, knowing where they break and how to build reliably on top of them.
• A track record of shipping LLM or generative AI applications used by real users in production; publications are valued less than shipped products.
• Ability to work from broadly defined goals, set your own objectives and drive work forward without a spec.
• High EQ and credibility with makers, able to earn the trust of a craft-focused team.
• Generous mentor, comfortable taking risks and learning from failure.

Required skills

• Hands-on experience writing and maintaining benchmarks and evaluation frameworks, ideally across more than one modality.
• Practical fine-tuning and model adaptation experience, with judgment on when not to use it.
• Sound judgment on inference trade-offs under tight compute, latency and cost constraints, including local or on-device deployment.
• Real software engineering depth: code others can maintain and things owned in production.

Nice to have: experience with agent frameworks, orchestration or evaluation tooling at production scale; game industry experience or a real appreciation for mobile; experience working alongside a platform or infrastructure team as an internal customer. Ideally based in Helsinki, though collaboration with other teams is required.

What we offer

• Relocation support from a dedicated mobility team and partners.
• No job titles with prefixes such as Senior, Junior, Principal or Director; recognition is not tied to title.