Research-Focused AI Engineer | PhD Experimental Physics
📧 Piotr.Gryko@gmail.com
🔗 LinkedIn • GitHub • GitLab
đź“Ť Europe
Professional Summary
Research-focused AI engineer with 12+ years experience bridging academic research and production systems. PhD in Experimental Physics with proven track record of identifying and correcting issues in published research, developing novel AI architectures, and scaling systems processing millions in monthly revenue. Specializes in document processing, computer vision, AI safety, and high-performance distributed systems.
Core Technical Expertise
AI/ML Research: Diffusion Models, Computer Vision, NER, LLM Safety, PyTorch, Signal Processing
Production Systems: Microservices, AsyncIO, Docker, High-Performance Computing
Data Engineering: PostgreSQL, Elasticsearch, Vector Databases, Real-time Pipelines
Languages: Python, C/C++, JavaScript/TypeScript, SQL
Infrastructure: Modal.com, AWS, Docker Swarm, CI/CD, Monitoring
Professional Experience
R&D Software Engineer | Hypodossier AG | Jun 2023 - Present
Document Processing Platform
- Architected extraction platform with microservices for PDF processing, text classification, and semantic analysis
- Developed multi-language document classification using SpaCy NLP for German financial documents
- Built asynchronous processing architecture using RabbitMQ and AsyncIO for high-throughput processing
- Client Impact: Serving 10+ banking clients with white-label solutions
AI Research Integration
- Collaborated with ML scientists to reproduce academic papers for product development
- Implemented NER using SpaCy and LayoutLLM for financial document analysis
Senior ML Engineer / AI Research Engineer | DevaLogic Projects | 2023-Present
Document Anonymization System
- Architected production-ready document anonymization using PyTorch, Diffusion Models (VAE/UNet), and NER
- Research Contribution: Corrected hyperparameter bugs in original DiffUTE paper, achieving significant performance improvements
- Built enterprise inference engine with thread-safe operations, LRU caching, batch optimization, and comprehensive OCR integration
- Designed scalable GPU training infrastructure on Modal.com (A100s) with W&B experiment tracking
- Impact: Featured in talks at EuroPython 2025 and PyCon Lithuania 2025
AI-Powered Document Chat Application
- Developed full-stack RAG system with self-hosted LLM components for enhanced privacy
- Implemented asynchronous Django REST API with streaming responses and WebSockets
- Built ChromaDB-based semantic search with LangChain text processing and SHA-256 deduplication
- Architecture: React TypeScript frontend, PostgreSQL backend, containerized deployment
Property Suitability ML Pipeline
- Developed end-to-end ML pipeline for energy efficiency assessment achieving F1 scores up to 0.78
- Engineered 40+ feature pipeline with geospatial data and NLP-derived insights using NLTK
- Implemented automated hyperparameter optimization using Optuna with MLflow tracking
Drone Dataset Collection System
- Built production AI pipeline for automated drone footage processing into YOLO training datasets
- Performance: AVideo processing with GPU acceleration and quality filtering
- Developed ensemble YOLO detection system with confidence thresholding and batch optimization
- Implemented microservices architecture with Kubernetes deployment and comprehensive monitoring
Property Energy Efficiency ML Pipeline | 2024 End-to-end ML system for EPC rating prediction
- Engineered 40+ features with geospatial and NLP components
- Achieved F1 score of 0.78 using XGBoost with Optuna optimization
- Built MLflow experiment tracking with automated hyperparameter tuning
Senior Software Engineer | Qogita | Mar 2022 - Jun 2023
Scaling MVP to $20M/month Revenue
- Led backend and data engineering operations using Django, Celery, Pandas, Snowflake, MongoDB
- Migrated payment system from Square to Wise with zero downtime
- Implemented bi-directional data synchronization between Snowflake and PostgreSQL
- Built pandas-based microservice for seller stockist data ingestion (Excel, CSV)
R&D Software Engineer | Unipart Digital | Oct 2015 - Jan 2022
Multidisciplinary Technology Leadership
- Led full-stack development projects combining web development, data analysis, and embedded systems
- Developed predictive analytics systems using Python, NumPy, Pandas for logistics optimization
- Managed small engineering teams (3-4 developers) delivering complex technical solutions
- Built production systems using Django, React, Docker with comprehensive DevOps practices
R&D Software Engineer | ION Geophysical | Jun 2013 - Oct 2015
High-Performance Computing Systems
- Developed signal processing tools for HPC systems using C++
- Implemented numerical algorithms optimized for geophysical data processing
- Integrated Python/NumPy into C++ processing systems for enhanced functionality
Education
PhD in Experimental Physics - Nanotechnology & Biomaterials | Imperial College London | 2007-2012
Research focus: Self-assembling biomaterials for bio-sensing applications, nanotechnology characterization
| MSci Physics, First Class Honours | University College London | 2003-2007 |
Recent Speaking Engagements
2025 Conferences:
“Anonymization of Sensitive Information in Financial Documents” - EuroPython 2025, PyCon Lithuania 2025, Data Science Summit: ML Edition 2025
2024 Conferences:
PyCon Lithuania 2024 - “Building and Scaling an AI Startup with Async Django” 🎥 Watch Talk
PyCon Lithuania 2024 - “Scaling, Refactoring and fixing a Django MVP for Production” 🎥 Watch Talk
Notable Projects & Research
AI Safety Research - Comprehensive LLM security testing suite for healthcare AI companions
Computer Vision Pipeline - 200-400 FPS autonomous drone detection system with quality filtering
Financial Document Intelligence - Semantic processing platform serving major European banks
Privacy-Preserving ML - Production document anonymization correcting published research flaws
Community Involvement
Founder & Organizer - “AI Code & Coffee Warsaw” tech mentoring meetup Facebook
Open Source Contributor - Active on GitHub and GitLab
Technical Mentor - Supporting Warsaw’s developer community through knowledge sharing
Publications & Research Impact
- Identified and corrected critical bugs in DiffUTE paper implementation (2024)
- Featured speaker at major European Python and AI conferences
- Research focus on practical applications of cutting-edge AI in regulated industries
- Bridge between academic innovation and commercial viability in AI systems
Available for consulting, speaking engagements, and research collaborations in AI/ML, computer vision, and privacy-preserving systems.