Fedor Konovalenko

Developed and deployed AI solutions across domains including cybersecurity, legal, mining, and manufacturing. Built LLM-powered assistants using RAG pipelines and optimized performance through prompt tuning and custom benchmarks. Delivered a mobile-ready OCR sys (ONNX) achieving 92% word recognition for Cyrillic documents. Designed real-time depth estimation for mobile (CoreML) and implemented CV-based productivity tracking, reducing error to <3%. Focus on end-to-end delivery from prototyping to production.

Delivered end-to-end CV solutions in medical imaging. Built a diagnostic system for bacterial sample analysis with HDR imaging, reducing analysis time by 30%. Trained microorganism recognition models and optimized post-processing for antibiotic resistance. Developed tumor detection models for CT scans and deployed them via FastAPI and Docker. Created a heart ultrasound analysis tool with instance segmentation and vital sign estimation, reducing review time by 60% through an interactive Gradio interface.

Led research and modeling efforts on complex physical systems, combining simulation, numerical methods, and experimental validation. Developed Python-based tools for solving differential equations and analyzing physical processes; utilized libraries like NumPy and Matplotlib for scientific computing and data visualization. Gained a strong foundation in applied mathematics, physics-informed modeling, and experimental design - skills that are now applied to ML model development and evaluation.