Data & Analytics

Pocket Worlds•Austin, TX

About The Position

This role owns the full analytics loop for a live-service game: frame the right question, get trustworthy numbers, build tools that make answers repeatable, and turn results into decisions the team acts on. It is part analyst, part analytics engineer, part internal-tools builder. AI coding agents do much of the hands-on work. The role supplies the judgment: what to measure, whether a result is plausible, and when something looks off and needs a second look.

Requirements

Strong SQL and Python for analysis
Applied statistics, with a practical sense of which methods suit noisy, skewed, real-world data and where they break down
Comfort building and deploying small services and data pipelines
Fluency with AI coding agents and the workflows around them
Notices when a number doesn't make sense and digs in before reporting it
Knows how to set up an analysis so it actually answers the question: comparing like with like, separating a real effect from noise or a shift in who's being measured, and recognizing when the data can't support a conclusion
Understands free-to-play games: monetization (including gacha), live events, virtual economies and how different player segments behave
Explains methods in plain language and writes for decision-makers: interpretation, not restated numbers

Nice To Haves

Familiarity with applied ML (embeddings, similarity, scoring) is a plus

Responsibilities

Own analytics across content releases, live events, monetization, the in-game economy and player behavior, from an ambiguous question to a clear recommendation.
Write complex SQL against large, messy, semi-structured event data.
Design and evaluate A/B tests, and reach sound conclusions when a clean experiment isn't possible.
Reconcile metrics across sources, and be the person who can explain why two numbers disagree.
Use AI agents as the main source of leverage for analysis, engineering and reporting.
Keep a critical eye on their output. Catch results that don't hold up, whether from flawed logic, bad data, or an analysis that quietly answers a different question than the one asked.
Build the shared context and processes that make AI-assisted analysis reliable, for yourself and for others.
Help less technical teammates answer their own questions safely.
Build bespoke internal dashboards and tools around the decisions people actually make, not generic reporting. Put the right answer in front of the right person, rather than a wall of charts everyone skims.
Treat adoption as the measure of success: a tool nobody uses is a failure, however well it's built. Work closely with users to learn what they need, including what they didn't know was possible, and keep iterating after launch.
Own these tools end to end, from data pipeline to interface to deployment to iteration.
Build and maintain the scheduled pipelines that feed them.
Build models and simulations as needed: forecasting, economy and event simulation, probability and expected-value tools, and ML-based content scoring.
Make tools usable beyond the data team with clear docs, APIs and integrations.
Turn findings into concrete recommendations for product, art, live ops and external partners, and see them through to shipped changes.
Own recurring performance reporting and target tracking, and explain what is driving results above or below plan.
Act as a thought partner on design and planning decisions, not just a source of numbers.