Jan Pecka

Background

About Jan Pecka

With more than six years of professional experience in data science and machine learning, I combine a solid mathematical background with a versatile skill set. I've worked across industries and data domains, bridging the gap between research, model development, and production - with a focus on real-word business impact. Throughout my career, I've worked on projects spanning multiple ML areas and covering the full ML cycle. I've built and maintained systems on GCP and Azure using tools like Databricks, Airflow, Kubernetes, and MLFlow. After starting on data analysis and engineering, I've worked extensively on recommender systems, unsupervised anomaly detection, and other classical ML problems. Recently, I’ve focused on deep learning for audio and vision, particularly with transformer architectures. I've collaborated on pre-training, fine-tuning and optimizing large models using frameworks such as TensorFlow, Torch and ONNX for edge deployment. Beyond implementation, I've led small projects as both a team lead and a technical owner, ensuring timely delivery aligned with business goals. I enjoy new challenges and diving deep into novel exciting problems. My focus is always on building reliable, well-engineered ML systems that follow the best coding practices and deliver clear business gains.