AI Research Engineer

The role

This role sits on a newly formed AI Research team at PostHog that is training its own deep learning models to power automated product insights. The engineer will help move ideas from research into production, working on problems like cheaper session replay analysis, predicting user drop-offs, and simulating users to test releases.

What you would do

• Explore promising modeling approaches and carry the most effective ones through to production systems.
• Improve the speed and cost of detecting problems in session recordings using stronger models and DOM mutation data.
• Build models that anticipate visitor drop-offs and suggest changes that could raise conversion without manual analysis.
• Develop simulated users that walk through product flows to catch confusing paths or breakages before release.
• Work close to customers to understand real needs and influence what the team ships.
• Write production-grade code and ship working features instead of stopping at prototypes.

What they asked for

• Strong grounding in mathematics, or prior hands-on experience training machine learning models.
• Hands-on proficiency with PyTorch.
• Strong coding ability in Rust, CUDA, or C, with Rust preferred.
• Working understanding of transformer architectures.
• A product mindset and a bias toward building things rather than only studying them.
• Ability to work hybrid from the London office.

Nice to have

• Doctoral training, with publication welcome but secondary to shipping.