Idler builds reinforcement learning environments that teach AI models to code like 0.01% engineers. Our training environments are based on real-world coding scenarios that frontier models will actually encounter.

We've closed a multimillion-dollar contract with a leading foundation lab (the largest they've issued to date), and demand is outpacing our capacity to deliver. We're scaling fast to build the training data layer for frontier AI. Every breakthrough model will need environments like ours to learn real-world skills at scale.

We're a tight-knit founding team in San Francisco. We move quickly, think deeply about what makes good training data, and play to win. Join us if you want to win too.

What We’re Building

We refine coding problems drawn from real engineering — not puzzles, but real reasoning from production code. Your job is to make the thinking clear: what changed, why it mattered, and how an engineer reasoned through it.

The Role (Contract)

We’re looking for engineers who can trace the intent behind a code change, understand what changes were made and why, and express that clearly in natural language.

This role is part-time, remote, and built for people with technical fluency, sharp attention to detail, and a disciplined command of English.

What You’ll Do

• Read unfamiliar repos and understand the intention behind commits

• Review and refine problem statements until they’re clear and accurate

• Adjust unit tests to keep problems fair and solvable

• Provide brief, consistent updates on progress and blockers

Who You Are

You have 1–2 years of software engineering experience and a precise eye for detail. You write cleanly, communicate quickly, and you see software engineering as your craft.

Requirements

• 1–2 years of software engineering experience

• Fluent English; strong writing is a must

• Responsive, independent, and attentive to detail

• 0–25 hours per week, flexible

How it Works

The best contributors stay and grow with us; they show care and depth in their work, communicate clearly about status and availability, and they help shape how our process evolves.

TypeScript, React, NextJS, Postgres, NodeJS, various AI completions