Data Scientist
Summary
Build and deploy ML models for mobile game LTV forecasting and UA spend optimization using Python, PyTorch, and XGBoost, collaborating with engineering and product teams.
About Kohort
Kohort is an AI-powered growth platform for mobile game studios, spanning user-acquisition optimisation, cohort-based LTV forecasting, growth financing, and M&A diligence. Our models are trained on over $6bn of UA spend across 300+ games and have underwritten $12bn+ in transactions. We are backed by AlbionVC, Eurazeo, The Raine Group, and Triple Point.
We are a group of gaming and data experts from King, Sensor Tower, Tripledot, Mistplay, Bloomberg, and Football Radar. Our clients range from large listed developers and publishers to up-and-coming start-ups building at the cutting edge of mobile gaming.
kohort.io
Why Kohort
- Career acceleration: We are accelerating as only a startup can. Join us in a zero-politics, no-drama, high-energy, impact-focused environment on the ground floor.
- Define your career: We are working on a $6bn problem — that's how much UA spend is wasted in mobile gaming each year. Let's fix it together.
- We live and breathe gaming: You'll work with the biggest game studios and on major gaming PE and VC deals.
What you will do
- Build, train, and improve machine learning models for LTV forecasting and ROAS prediction.
- Work with large-scale cohort and user-acquisition data to extract reliable signals.
- Help ship models into production pipelines that drive daily, automated campaign recommendations.
- Collaborate closely with engineering and product to turn models into customer-facing features.
Experience
- Currently working as a Data Scientist, ML Engineer, AI Engineer, or Software Engineer
- Solid Machine Learning background — model training, evaluation, and iteration (essential).
- Strong Python and SQL.
- Hands-on experience with: PyTorch (torch), XGBoost, numpy, polars, git.
Location
- Based in South Africa (Cape Town preferred). Remote-first, with access to in-office collaboration and several company-wide gatherings throughout the year.
Preferred
- A STEM degree — engineering, computer science, or mathematical statistics — ideally from UCT, Wits, Stellenbosch University, or the University of Pretoria.
- Exposure to production ML / MLOps, time-series or forecasting problems, or large-scale data.
- Experience with distributed, batch, or streaming data frameworks.
- Interest in gaming or ad-tech.
What you will receive
- Competitive salary
- Employee Share Scheme
- Income protection and life cover
- Wellbeing rewards
- Employee Assistance Program