Principal Competitive CPU Performance Forecaster & Data Scientist

ADVANCE YOUR CAREER. ADVANCE THE WORLD.  At AMD, we believe technology has the power to solve the world’s most important challenges. From advancing healthcare and scientific discovery to powering AI and the technologies people rely on every day, innovation at AMD is shaping the future.    Whether you’re designing next-gen processors, enabling AI breakthroughs, or bringing leading edge products to market, every role at AMD contributes to something bigger — technology that moves the world forward. Join us and, together, we’ll advance your career.

The Competitive Advanced Performance (CAP) team is a new, visible capability focused on predicting, validating and explaining competitive server CPU performance before products reach the market. Joining now means helping define the methods, tools and operating model from the ground up.

AMD is seeking a data scientist and performance engineer to own the quantitative forecasting system behind the CAP team. You will combine architecture assumptions, workload measurements, software trends, platform data and partner validation into forward-looking predictions with explicit confidence ranges. The role is accountable for model error, scenario analysis, forecast-versus-actual learning and a clear competitive narrative: where AMD is likely to lead, where it may face risk, how confident the team is and which variables could change the outcome.   THE PERSON:   You are neither a pure data scientist who treats performance data as opaque features nor a benchmark engineer who reports only point results. You understand enough CPU and system performance to model causality, and enough statistics to avoid false precision. You care about provenance, versioning and calibration, and you can make uncertainty understandable to engineers and executives.

Design and maintain a forecasting framework that links CPU architecture, memory, I/O, power, operating-system and runtime behavior, and workload kernels. Create performance, performance-per-watt and price-performance forecasts with confidence ranges rather than single-point estimates. Build scenarios for uncertain competitor attributes such as frequency, core count, memory bandwidth, software maturity, launch timing and platform power. Track prediction error by workload, competitor, forecast horizon and model version; decompose errors and tune the model as measured systems become available. Develop normalized competitive scorecards across workload performance, efficiency and TCO, memory and I/O, software ecosystem, deployability and roadmap credibility. Fuse benchmark results, performance counters, partner telemetry, public roadmaps, software changes and platform evidence while preserving source provenance. Identify leading indicators that materially change the forecast and surface early risk or opportunity signals. Partner with architecture and workload leads to design experiments that reduce the most valuable uncertainties. Build executive-grade visualizations of deltas, confidence ranges, scenarios, drivers and forecast-versus-actual history. Write quarterly forecast narratives and support rapid recalibration when new silicon or material evidence appears.

Demonstrated experience building quantitative models used for technical or business decisions under uncertainty. Strong programming and data-analysis skills in Python or an equivalent analytical environment. Expertise in defining meaningful error metrics, calibration methods and sensitivity analyses for sparse or biased data. Working knowledge of server CPU and system performance and a willingness to engage deeply with architectural causality. Experience with reproducible data pipelines, versioning, notebooks or scripts, and source provenance. Strong visualization, writing and presentation skills for technical and executive audiences. Sound judgment about when a model is useful, when it is overfit