# Piotr Semberecki [@piotr\-semberecki](https://hacklab.so/piotr-semberecki) Piotr Semberecki is Senior Robotics AI and Simulation Engineer at SoftServe, based in Warsaw. No AI coding activity was synced in the last 30 days. Robotics AI engineer at SoftServe building robot\-learning, simulation and computer\-vision systems with Python, C\+\+, RL and imitation learning\. Location: Warsaw Joined: 2026-09-07 0 followers · 0 following ## Links - [Luma](https://luma.com/user/usr-rOmNZCAKjusjkuA) - [LinkedIn](https://linkedin.com/in/piotr-semberecki) ## AI usage (All time) No activity in this period. ## Accomplishments ### [Best hardware project: Quadcopter — Team Project Conference 2013](https://www.linkedin.com/in/piotr-semberecki/) Profile lists a best\-hardware\-project award for Quadcopter at the 2013 Team Project Conference\. ### [First place — Multiple Travelling Salesman Problem Competition](https://www.linkedin.com/in/piotr-semberecki/) Profile lists a diploma for first place in a Multiple Travelling Salesman Problem Competition; the award date is not supplied\. ### [Scheduling\-related patent co\-inventor](https://www.linkedin.com/in/piotr-semberecki/) In the Tooploox role description, reports being named as a co\-inventor on a patent related to reinforcement\-learning scheduling for refinery tanks\. The export does not provide the patent number or an independent registry link\. ### [Plugin networks for inference under partial evidence](https://www.linkedin.com/in/piotr-semberecki/) Research publication listed in the LinkedIn export\. Venue, publication date and individual contribution are not specified in the PDF\. ### [Distributed classification of text documents on Apache Spark platform](https://www.linkedin.com/in/piotr-semberecki/) Research publication listed in the LinkedIn export\. Venue, publication date and individual contribution are not specified in the PDF\. ### [Deep Learning Methods in Reinforcement Learning: Continual Learning and Explainable AI in Robotics Applications](https://www.linkedin.com/in/piotr-semberecki/) Research publication listed in the LinkedIn export\. Venue, publication date and individual contribution are not specified in the PDF\. ### [Deep learning methods for subject text classification of articles](https://www.linkedin.com/in/piotr-semberecki/) Research publication listed in the LinkedIn export\. Venue, publication date and individual contribution are not specified in the PDF\. ### [Multi\-agent Reinforcement Learning in Real\-time Strategy Games](https://www.linkedin.com/in/piotr-semberecki/) Research publication listed in the LinkedIn export\. Venue, publication date and individual contribution are not specified in the PDF\. ### [Patent listed: Directional stream value analysis system and server](https://www.linkedin.com/in/piotr-semberecki/) Title appears twice in the patent sidebar and is retained once\. The PDF does not specify patent number, jurisdiction, inventor list or legal status\. ## Experience ### Senior Robotics AI and Simulation Engineer — SoftServe 2026\-08 – Present Designs and builds intelligent systems for mobile robots and robotic arms across simulated and real\-world environments\. Work spans foundation models, reinforcement and imitation learning, synthetic\-data generation and computer vision in NVIDIA Isaac Sim, including NVIDIA Newton Physics engine optimisations\. ### Senior Machine Learning Engineer — GlobalLogic 2026\-02 – 2026\-07 Wrocław, Poland Developed agentic generative\-AI solutions for document\-correctness validation and semantic text matching\. The profile also lists time\-series anomaly detection for cellular networks\. ### Machine Learning Engineer — Contract / Freelance 2025\-11 – 2026\-07 Contracted with corporations and startups on generative AI and agents\. Developed reinforcement\-learning environments that simulate agent interactions, task objectives and performance\-feedback loops; built environments for software\-development agents with automated task generation and evaluation\. Experimented with local deployment of DeepSeek, Kimi and Qwen for applications including robotics\. ### Senior AI Software Engineer — SpyroSoft 2019\-06 – 2025\-12 Led solutions or contributed to multidisciplinary ML projects across computer vision and robotics\. Projects included traffic\-infrastructure data processing for automotive models; deep\-learning building segmentation from satellite imagery; real\-time depth\-based obstacle detection for autonomous cleaning robots; and radar classification of road users using Pandas, NumPy and scikit\-learn, followed by C\+\+ optimisation\. Built road\-lane detection from 2D/3D point clouds, with clustering/polynomial fitting and OpenDRIVE output\. Developed convolutional/recurrent gesture\-recognition models, GPU training and TensorFlow Lite deployment on Android\. Designed edge\-AI/GPU performance\-test pipelines using TensorFlow, PyTorch and CUDA C\+\+ on NVIDIA Xavier\. ### Machine Learning and Computer Vision Engineer — Tooploox 2017\-11 – 2019\-05 Worked on ML/computer\-vision solutions, GAN/custom\-architecture research, client technical discussions and business proposals\. Developed reinforcement\-learning algorithms for refinery tank\-scheduling problems, contributing to architecture and testing and being named as a patent co\-inventor\. Built bin\-picking automation using KUKA robots, Intel RealSense cameras, Python and ROS; explored industrial deep\-learning/anomaly\-detection applications; developed anomaly\-detection software for a three\-depth\-camera fitness/body\-volume device; and built clothing\-item detection and visual\-style recommendation systems\. ### Graduate Research and Teaching Assistant — Wrocław University of Science and Technology 2016\-11 – 2017\-07 Wrocław, Poland Conducted doctoral research in machine learning, deep neural networks, NLP and computer vision and prepared research\-grant proposals\. Taught Python development through telecommunications\-channel simulation, covering Forward Error Correction, Automatic Repeat Requests, Hamming codes and scrambling\. ### Embedded Software Engineer — Nokia Networks 2014\-07 – 2016\-09 Wrocław, Poland Developed Python, MATLAB and C\+\+ software for LTE physical\-layer component testing, including framework design, test suites and Python–MATLAB pipelines\. During an internal gamification project, built Python backend and Angular 2 frontend software\. ### Student Position — NeuroSoft PL 2013\-07 – 2013\-08 Wrocław, Poland Developed an FPGA image\-processing system over TCP/IP using VHDL and Verilog\. Evaluated Ethernet/GigE image\-processing performance through a VHDL histogram\-equalisation implementation\. ## Education ### Doctor of Philosophy \(PhD\), Computer Science — Wrocław University of Science and Technology 2015–2026 \(year\-only period listed in the export\)\. ### MScEng, Control Engineering and Robotics — Wrocław University of Science and Technology 2014–2016 \(year\-only period listed in the export\)\. ### MScEng, Computer Science — Wrocław University of Technology 2014–2015 \(year\-only period listed in the export\)\. ### BSc Engineering, Computer Science — Wrocław University of Technology 2010–2014 \(year\-only period listed in the export\)\. ### BSc Engineering, Control Engineering and Robotics — Wrocław University of Technology 2008–2012 \(year\-only period listed in the export\)\.