Planning & Capacity Engineer at Meta. I own demand planning and supply planning for the server racks and critical components behind hyperscale AI compute infrastructure, turning fast-changing GPU and accelerator demand into executable hardware delivery plans, supplier commitments, and a dependable supply signal the rest of the organization plans against.
PhD physicist and supply chain optimization leader with a decade of experience in demand forecasting, capacity planning, S&OP, inventory and buffer strategy, and large-scale mathematical optimization. As a consultant at Accenture and End-to-End Analytics I delivered forecasting, procurement optimization, configuration optimization, and GenAI decision platforms for automotive, high-tech, and retail enterprises that lifted forecast accuracy, reduced buffer inventory, and unlocked tens of millions of dollars in annual value.
Hands-on, AI-native engineer: I build the data pipelines, forecast validation, automation, and internal planning tools that make hyperscale infrastructure planning trustworthy at scale, on cloud-native, serverless architectures (AWS, Azure, GCP), with Python, SQL, Spark, and optimization solvers such as Gurobi, CPLEX, and OR-Tools.
Based in the SF Bay Area. Open to conversations about AI infrastructure supply chain, capacity planning, demand planning, optimization, and decision platforms, especially where speed to value and scale matter.
Professional Experience
2026–now
Planning & Capacity Engineer
Meta
Infrastructure supply chain · Menlo Park, CA
2022–2026
Senior Manager
Accenture
AI & analytics programs across automotive, high-tech and retail
2021–2022
Manager
Accenture
via acquisition of End-to-End Analytics
2018–2021
Data Science Manager
End-to-End Analytics
Palo Alto, CA
2017–2018
Senior Data Scientist
Acorn Analytics
San Francisco Bay Area
2017–2018
Assistant Professor
University of Warsaw
Faculty of Physics
2012–2016
Researcher and PhD candidate
University of Warsaw
2009–2022
Entrepreneur
iStudio
side venture
Selected & Recent Projects
AI compute hardware supply chain planning at Meta
2026 – now
Translate the explosive growth in AI compute demand into executable hardware fulfillment plans, supplier commitments, and delivery scenarios across racks and critical components
Own demand and supply planning with the world's largest ODM, one of the largest hardware supply relationships in the industry, converting their capacity commitments into the reliable supply signal downstream capacity and product teams plan against
Build the data foundation and internal tooling behind it: forecast validation, automation, and data pipelines
Help shape a next-generation, in-house planning platform spanning demand planning, forecast generation, and allocation
Cloud-native decision platforms for automotive, high-tech and retail clients (Accenture)
2022 – 2026
Forecasting and procurement optimization engine on AWS with Gurobi that improved forecast precision by 25%+ and cut buffer stock by 18%
Vehicle configuration optimization capability on Azure using advanced constraint programming
Secure, serverless entity-search application with fuzzy matching that reduced manual review by 90%+
Reusable GenAI accelerators for survey intelligence and market opportunity discovery (Azure OpenAI / OpenAI), turning unstructured text into executive-ready insights
Owned product strategy, architecture, and delivery; partnered with executives to align roadmaps to measurable outcomes
Advanced planning and optimization for enterprise clients (Accenture)
2021 – 2022
Led architecture and core constraint logic for a dealer ordering recommendation engine for a US market-leading car manufacturer, forecasting demand for fully configured vehicles (Python, CPLEX, Gurobi, or-tools)
Built Spark / Hadoop ETL at scale and optimized runtime on a cluster with thousands of cores, driving an estimated $70M+ of annual value
Conducted a supply and demand planning capability assessment for an emerging mobility manufacturer; benchmarked Anaplan, SAP, o9 and Blue Yonder and produced a migration roadmap and change plan
Ran a hardware demand forecasting gap analysis for a hyperscale cloud provider; mapped the planning lifecycle, designed a short/mid-term optimization prototype, and recommended algorithm and metric improvements
Analytics and optimization for Fortune 500 supply chains (End-to-End Analytics)
2018 – 2021
Built a cloud-native vehicle routing and inventory replenishment engine (AWS Lambda, Fargate, S3, EC2) for a nationwide car battery supplier, reducing delivery delays by 25% while maintaining fill rates
Developed a large-scale device telemetry analytics platform for a Fortune 500 phone manufacturer to diagnose battery and charging issues across a global fleet, enabling faster triage and lower warranty cost
Managed development of a .NET Core promotion-management platform for a North American bakery leader, optimizing $1.5B in trade spend
Architected the AWS infrastructure, resilient nightly ETL processing tens of gigabytes, CI/CD (GitHub, Jenkins, Lambda, Fargate), and QA tooling behind it
Drove stakeholder alignment and transitioned prototypes into production systems
Streaming IoT analytics for a wearable device maker (Acorn Analytics)
2017 – 2018
Built statistical models and serverless pipelines (AWS Lambda + S3) to detect user behavior patterns, device performance issues, and usage-based triggers on live device data
Reduced support response time by 35% and informed the product roadmap with quantitative insights used in designing the next device iteration
Partnered with product, data engineering, and customer success to define KPIs, instrumentation, and experiments
Photos
Open Source
glucose-syrup-modifiedModified Glucose SAT Solver for purpose of finding all feasible solutions (and solution groups) for given SAT problem.
pySeismicFMMPython based travel time calculation in regular 2D and 3D grids in Cartesian and geographic coordinates using Fast Marching Method
pyMultiMinePython based cryptocurrency mining driver that allows switching between different currencies, mining software and algorithms based on current mining profitability.
Publications
17 peer-reviewed papers in geophysics and seismology.
2024
Karkowska, K., Wilde-Piórko, M., Dykowski, P., Sękowski, M., Polkowski, M.
SM7.1 New developments in open source and academic seismic software
European Geosciences Union, General Assembly 2017
Primary Convener: Marcin Polkowski · Conveners: Mariusz Majdański, Ilma Janutyte
2016
S020: New Developments in Open Source and Academic Seismic Software
American Geophysical Union, Fall Meeting 2016
Primary Convener: Marcin Polkowski · Conveners: Mariusz Majdański, Monika Wilde-Piórko
2016
SM7.5 New developments in open source and academic seismic software · SM1.7 Uncertainty estimation in geophysical methods
European Geosciences Union, General Assembly 2016
Primary Convener: Marcin Polkowski (SM7.5) · Conveners: Ilma Janutyte, Mariusz Majdański, Richard Hobbs
2015
S33C: New Developments in Open Source and Academic Seismic Software Posters
American Geophysical Union, Fall Meeting 2015
Primary Convener: Marcin Polkowski · Convener: Mariusz Majdański
Presentations & posters
2017
EGU General Assembly 2017
23–28 April 2017, Vienna, Austria
pySeismicDQA: open source post experiment data quality assessment and processing PDF
2016
AGU Fall Meeting 2016
14–18 December 2016, San Francisco, USA
pySeismicFMM: Python based travel time calculation in regular 2D and 3D grids in Cartesian and geographic coordinates using Fast Marching Method PDF
2016
EGU General Assembly 2016
17–22 April 2016, Vienna, Austria
pySeismicFMM: Python based travel time calculation in regular 2D and 3D grids PDF
Broadband network on-line data acquisition system with web based interface for control and basic analysis PDF
2015
AGU Fall Meeting 2015
14–18 December 2015, San Francisco, USA
Travel time calculation in regular 3D grid in local and regional scale using fast marching method PDF
Unexpected non seismic signals recorded by broadband seismic stations during geomagnetic storms
2015
EGU General Assembly 2015
12–17 April 2015, Vienna, Austria
Applications of detailed 3D P-wave velocity crustal model in Poland for local, regional and global seismic tomography PDF
2014
AGU Fall Meeting 2014
15–19 December 2014, San Francisco, USA
High-Resolution 3D P-Wave Velocity Model in the Trans-European Suture Zone in Poland PDF
Passive Seismic Experiment “13 BB Star” in the Margin of the East European Craton, Northern Poland
2014
EGU General Assembly 2014
27 April – 2 May 2014, Vienna, Austria
“13 BB star” — broadband seismic array at the edge of East European Craton in Poland (oral)
Local seismic events in area of Poland based on data from PASSEQ 2006-2008 experiment PDF
2013
EGU General Assembly 2013
7–13 April 2013, Vienna, Austria
Relations between heat flow, topography and Moho depth for Europe PDF
2012
ESC 33rd General Assembly
19–24 August 2012, Moscow, Russia
Seismic wave velocities in the sedimentary cover of Poland: Borehole data compilation PDF
Teaching
At the University of Warsaw I taught computer science and seismic modeling and advised student projects. Every semester I had one or two groups of students; depending on the semester I taught Python programming, SQL and database programming, microcontroller programming or basic electronics.