I need an experienced cricket data scientist to turn years of ball-by-ball history into two working assets: a first-innings score predictor and a live match-winner predictor, both tuned exclusively for T20 cricket. Test and ODI knowledge is welcome, but every decision you make has to serve the fast-paced T20 format. You should already be comfortable wrangling sports data with Python (Pandas, NumPy, Scikit-learn, or similar), feature-engineering cricket-specific variables, and pushing trained models into a low-latency environment so the predictions can be consumed in real time. If you have deployed models for live IPL dashboards or betting APIs before, tell me about it—your relevant experience is what I will weigh most heavily when selecting the freelancer. The ideal flow I imagine is • Clean and enrich the historical T20 datasets. • Build, test, and validate the two models. • Expose them through a lightweight API or similar mechanism ready for production. Because the project is time-sensitive, I would like to see an initial milestone—even a proof-of-concept notebook—and a production-ready solution as soon as possible thereafter. When you reply, focus on your direct experience with cricket or comparable sports analytics, model deployment track record, and any T20-specific insights you have used before. I am ready to start immediately and will move quickly once I see the right fit.