MARKUS PLANDOWSKI

I build end-to-end ML systems for real-world sensor data.

ML Engineer & Consultant 04/2024–Present | Self-employed | Germany (Remote)

Built batgrad, a public, reproducible template that takes heterogeneous battery data from ingestion through training to long-horizon prediction.

  • Built a reproducible research pipeline that carries traceable cycling datasets through Optuna-based architecture search across multiple GPUs to rollout-validated checkpoints.
  • Used gradient diagnostics to identify training instabilities and guide changes to model architecture and learning objectives.
  • Redesigned validation around long-horizon rollouts that reflect autoregressive inference rather than next-step prediction alone.

Separately, completed a seven-month consulting engagement with Chemix Inc., a Silicon Valley battery startup.

  • Introduced distributed hyperparameter optimization using SMAC3, Dask, and Weights & Biases, identifying a more parameter-efficient model with improved validation performance.
  • Built reusable, standardized cycling workflows spanning PyBaMM-based long-horizon synthetic data generation and laboratory test-cycler execution.

10/2023–03/2024 | Solo Motorcycle Travel | Vietnam

ML Research Intern & Master's Thesis 06/2022–09/2023 | Fraunhofer ISE | Freiburg, Germany

Proposed and conducted a study comparing ML models for state-of-charge estimation in NMC and LFP cells. Systematic benchmarking and hyperparameter optimization favored a decoder-only Transformer as the most practical neural approach, while validated cell-specific equivalent-circuit models remained more accurate on measured data.

Embedded Systems Intern 06/2021–05/2022 | Telocate GmbH | Freiburg, Germany

Designed and built an end-to-end smart-glove prototype with a custom acquisition PCB, an FPGA interface, and an ML model that detected object presence in real time.

10/2018–10/2020 | Travel & Licensed Electrical Work | Asia and Australia

Battery Engineer 04/2017–09/2018 | MAHLE GmbH | Stuttgart, Germany

Owned electrical interface design for an 800 V/750 kW immersion-cooled battery pack prototype.

  • Drove cross-functional electrical integration from module development through full-pack validation, resolving competing technical requirements across internal departments, suppliers, and OEM stakeholders to deliver a prototype that met its specifications.
  • Developed a reusable transient electrothermal busbar simulation tool that replaced repeated one-off analyses and accelerated hotspot assessment during design iterations.
  • Named co-inventor on two battery-system patent publications: US20200153060A1 and US20200119415A1.

Hardware Intern & Bachelor's Thesis 04/2016–03/2017 | MAHLE GmbH | Stuttgart, Germany

Brought up and validated a bidirectional DC/DC converter for thermoelectric waste-heat recovery through driver and sensor characterization, SPICE-based fault analysis, bench measurements, and vehicle-system modeling in Simulink and PLECS.

M.Sc. Embedded Systems Engineering 11/2020–09/2023 | University of Freiburg | Freiburg, Germany

B.Eng. Mechatronics 10/2012–03/2017 | Ulm University of Applied Sciences | Ulm, Germany

Mechatronics Apprenticeship & Licensed Electrician 08/2008–09/2015 | OSRAM AG | Augsburg, Germany

Returned during semester breaks as a full-time licensed electrician, working on switchboard wiring, high-power cabling, and fault tracing.