Robotics & Physical AI Engineer
Uncertainty-aware reinforcement learning · Imitation learning · Sim-to-real transfer
MSc student in Trustworthy AI Systems at Wrocław University of Science and Technology. Junior Physical AI Engineer at xBerry. Background in automation and control engineering.
I work on one question:
When should a learned policy not be trusted, and can that be measured at runtime?
Uncertainty-Aware Surrogate-Assisted Evolutionary Reinforcement Learning in Expensive Low-Data Regimes
ECML PKDD LFSD 2026 Workshop
Population gating driven by a surrogate that estimates uncertainty over critic value predictions.
Do Edges Save Energy? Classical Preprocessing for Efficient ACT-Based Robot Learning
ISW 2026
Ablation of visual preprocessing for Action Chunking Transformers.
SAR Land Cover Classification: Do Heavyweight Deep Learning Models Justify Their Cost Against Lightweight Architecture?
EUSAR 2026
Memory-efficient deep learning for synthetic aperture radar data.
| Languages | Python · C++ |
| Deep learning | PyTorch · JAX · Hugging Face (LeRobot, Transformers, Diffusers) |
| Simulation | NVIDIA Isaac Sim · Isaac Lab · Newton · MuJoCo / MJX · Gymnasium |
| Deployment | ONNX Runtime · TensorRT · Docker |
| Tools | Hydra · Weights & Biases · MLflow · uv · Ruff · pytest |
| Agentic AI | LangChain · LangGraph |