Lead AI Engineer, Game Generation — Text-to-Game Pipeline

TechTree; Greater London; On-site; GBP 120,000 - 180,000; Full time

Job summary

TechTree is assisting a client in London to hire a Lead AI Engineer, Game Generation. This role is remote until Oct 2026, after which relocation to London is required.

You will own the end-to-end text-to-game pipeline—from prompt extraction to running, playable output, at scale. You will drive cost and performance improvements, run benchmark suites, and implement routing, caching and distillation to reduce inference costs.

Qualifications

• Shipping text-to-game or prompt-to-running-app tech at scale.
• Own the generation → validation → repair → eval loop with real pass-rate and cost numbers.
• Strong runtime engineering skills in WebGL-based libraries (Three.js, Phaser, Babylon).

Responsibilities

• Own the full pipeline — intent extraction, model routing, generation, validation, repair, runtime.
• Move playability pass rate up and cost per generation down, at volume.
• Run benchmark suites and LLM-as-judge evals, and cut inference cost through routing, caching and distillation.

Skills

Three.js

Phaser

Babylon

Procedural generation

Tools

Three.js

Phaser

Babylon

Job description

This is a role that TechTree is recruiting for on behalf of one of our clients.

TechTree is an AI-driven recruitment platform working with high-growth companies.

When you apply, our AI Agent matches you not just to this role, but to other relevant opportunities across our network, so one application can unlock multiple roles.

Lead AI Engineer, Game Generation — own a text-to-game pipeline running millions of generations. London (remote until Oct 2026, then relocation).

Our client runs a mobile feed made entirely of AI-generated browser games created from a prompt. The generation pipeline is the product.

This is the technical owner role for that pipeline — from prompt to running, playable game, at scale and at a cost that works.

What You'll Do

• Own the full pipeline — intent extraction, model routing, generation, validation, repair, runtime


• Move playability pass rate up and cost per generation down, at volume


• Run benchmark suites and LLM-as-judge evals, and cut inference cost through routing, caching and distillation

What We're Looking For

• Have shipped text-to-game or prompt-to-running-app tech at scale (Rosebud, Replit Agent, Bolt, v0 calibre)


• Own the generation → validation → repair → eval loop, with real pass-rate and cost numbers


• Strong enough engineer to live in the runtime — Three.js, Phaser, Babylon, procedural generation

Comp is open — top of market for the right person.

What is TechTree?

TechTree combines AI agents with a human support team to give candidates a better, more transparent hiring experience.

Our AI Agent

gets to know your background, experience, and what you're optimising for. It continuously matches you to live roles across your network where there's genuine alignment, not just keyword overlap.

Our Human Team

is there to support you through the process — answering questions, adding context, and making sure you're never left guessing or ghosted.

We only post real jobs.

Every role on TechTree is a live, active hire.