I’m a research-focused AI engineer who bridges the gap between cutting-edge academic research and production systems. With a PhD in Experimental Physics from Imperial College London and 12 years of engineering experience, I specialize in translating complex theoretical concepts into scalable, real-world AI applications.
My Expertise spans deep learning, computer vision document processing, and full-stack development. I don’t just implement existing models—I research, debug, and improve upon published academic work, often identifying and correcting critical issues in original research implementations to achieve production-ready performance. International speaker on AI/ML topics at major Python conferences.
Core Competencies AI/ML: PyTorch, Diffusion Models, YOLO, NER, LLMs (LLAMA, GPT), RAG Systems, Computer Vision Backend: Python, Django, FastAPI, Async Programming, Microservices, REST APIs Infrastructure: Docker, RabbitMQ, PostgreSQL, Redis, AWS, Azure, Modal.com, CI/CD, Gitlab Data: Pandas, NumPy, Elasticsearch, Vector Databases, Snowflake, ETL Pipelines
Research & Innovation
My recent focus areas demonstrate deep technical expertise:
- Diffusion Models Research: Reimplemented DiffUTE paper implementation, achieving significant performance improvements for document anonymization tasks (fixed hyperparameter bug and )
- Computer Vision Systems: Developed ensemble YOLO detection pipeline for autonomous drone dataset creation, with scene detection and quality filtering algorithms
- AI Safety & Security: Built comprehensive testing frameworks evaluating LLM robustness against prompt injection attacks and medical question misclassification across multiple model architectures
- Document Intelligence: Built semantic document processing combining NER, OCR, and machine learning for automated categorization of complex financial documents
Technical Architecture
Beyond algorithms, I architect production systems that scale:
- Distributed Training: Cloud-native GPU training infrastructure, experiment tracking with Weights & Biases
- High-Performance Computing: Async-first microservices with GPU acceleration, batch optimization, and memory-efficient processing
- Research Infrastructure: End-to-end ML pipelines with automated hyperparameter optimization, comprehensive testing frameworks (80%+ coverage), and reproducible experiment workflows
Academic Rigor in Industry
My PhD background in experimental physics provides a unique analytical approach to engineering challenges. I approach each problem with scientific methodology—forming hypotheses, designing controlled experiments, and validating results through rigorous testing. This research mindset enables me to:
- Identify and resolve fundamental issues in published research implementations
- Design novel approaches to complex AI problems in financial and defense sectors
- Bridge the gap between academic innovation and commercial viability
Community Leadership
As a technical leader, I contribute to advancing the field through conference talks at EuroPython and PyCon Lithuania, focusing on practical applications of advanced AI techniques. I mentor developers through “AI Code & Coffee Warsaw” and believe in sharing knowledge to elevate the entire community’s technical capabilities.
My goal is to push the boundaries of what’s possible with AI while ensuring these advances translate into robust, ethical, and scalable solutions that solve meaningful real-world problems.