AI Research Engineer

• Join PostHog's brand new AI Research team, training deep ML models on petabytes of product data — events, errors, session replays, and revenue data — to surface insights and suggest fixes automatically. • Work on production-grade AI research across session replay analysis, user behaviour prediction, and synthetic user testing. • A unique role combining research and shipping: you will publish, but shipping to production has priority. • Real autonomy, close customer proximity, and significant influence over what actually gets built. • Hybrid role based at the PostHog London office.

• Strong background in mathematics — a hard requirement regardless of academic background. • Strong PyTorch experience. • Strong ability to code in Rust, CUDA, or C (Rust preferred). • Solid understanding of Transformers. • A product mindset — you will be building things, not just researching. • Ability to move from research to production; this is not a Jupyter notebook role.

• A PhD in a relevant field. • Prior experience training ML models at scale. • Familiarity with product analytics, session replay, or behavioural data systems.

• Research and productionise deep ML models using PostHog's petabyte-scale dataset of events, errors, session replays, and revenue data. • Work on session replay analysis to make issue detection faster, cheaper, and best-in-class. • Develop user behaviour prediction models to identify drop-offs and suggest improvements automatically. • Explore synthetic user testing to simulate users and catch confusing flows or breakages before release. • Ship research into production — publishing is welcome but secondary to shipping.

We have used PostHog's web analytics ourselves and genuinely loved the experience. The product is fast, thoughtfully designed, and surprisingly fun to use. PostHog is unlike most SaaS companies: their entire handbook is public, their codebase is open source, and they share revenue figures and board meeting notes with every team member. The company has raised $194M from investors, including Stripe, GV, and Y Combinator, reached a $1.4 billion valuation in 2025, and is targeting $100M in ARR by the end of 2026. They pay generously, benchmarking engineering salaries at the 90th percentile of San Francisco rates adjusted by location, with a fully public compensation calculator so you know your salary before you even apply. Employees on Glassdoor consistently highlight the flexible spending, genuine remote-first culture, and regular offsites, alongside a $1,500 quarterly budget for meeting up with colleagues anywhere in the world. Many team members are former founders, and the company is structured around small, autonomous teams that own their products end-to-end. If you are an engineer or builder who thrives on autonomy, ships fast, and wants to work on a product used by nearly 200,000 teams, PostHog is a rare opportunity.

• Salary of £128,999–£141,220 (Birmingham benchmark; may vary by location and step) plus significant equity. • Unlimited vacation with a minimum enforced. • Two meeting-free days per week (Tuesdays and Thursdays). • Home office budget and coworking credit. • Private health, dental, and vision insurance. • Training budget and company offsites. • Pension contributions.

PostHog is the only platform that acts as a co-pilot for product and AI agent development — shipping product analytics, a built-in data warehouse, session replay, and AI-powered analysis from a single platform. Launched from Y Combinator W20, PostHog has more than 450,000 organisations installed, is default alive, and has raised over $180m. It tracks more products across more dimensions than any other company on earth.