Data Science Intern, 4 weeks, Counter-Strike 2 Knowledge Required

Remote. 4 weeks, flexible start. Full or part time, no hour tracking. $3,000 USD flat stipend.

PureSkill.gg is an automated coach for Counter-Strike 2. We parse players' demos and generate personalized feedback with no human in the loop.

The project

Build a SOTA round win probability model. Given the state of a round at any tick (economy, positions, how many players are alive, equipment, bomb state, time), predict the odds each side wins it. Player positioning will be an important part of the model.

Additionally, every prediction has to be explainable. At any tick we need to know which factors moved the probability and by how much. A number nobody can interpret is useless for coaching, so you will trade some accuracy for interpretability and make the case for where you landed.

You would explore our tick-level data, build features, train and validate a model, build the explanation layer, and write up what you found.

Requirements

Counter-Strike. You need real game sense, because that is where the features come from. We will ask you about the game in the interview.

Python. Part of the interview is unassisted, so the fundamentals need to be yours. The other part is open to any tools you want to use.

Study. You are studying or have studied data science, computer science, statistics, machine learning, or similar.

Helpful but not required: XGBoost or similar, SHAP or other attribution methods, any sports or esports analytics work.

Non-traditional paths are welcome. If you are transitioning out of the military, changing careers, or got here later than most, apply. What matters is the three things above, not how straight the line to them was.

This model will be shipping to production!

Apply

Email with your CS2 background (rank and hours) and your resume. Skip the cover letter.

APPLICATIONS CLOSE AUGUST 3