UPscale  ·  Featured Work  ·  Stack  ·  Education  ·  Contact

Computer Science B.Sc. graduate with a specialization in Deep Learning, and an M.Sc. in Computer Science student at Ben-Gurion University of the Negev.

I work on Agentic AI and LLM systems, Deep Learning and Computer Vision — and I build them as production systems. That means the whole path: data and degradation modeling, architecture and training, evaluation harnesses, then serving it behind a real API with persistence, tool calling and retrieval underneath.

Recurrent video super-resolution built from the ground up in PyTorch: a BasicVSR-style network with SPyNet optical-flow alignment and bidirectional recurrent feature propagation over a 15-frame temporal window, trained on REDS with synthetic degradation modeling and fine-tuned on H.264 codec-aware degradations (V4) to close the gap to real-world compressed footage.

_{Real 256×144 input · real 1024×576 model output — watch the full clip in HD}

_{Evaluated on REDS4 over 430 evaluation windows. V3 → V4 codec fine-tuning adds +0.36 dB on compressed video.}

Stack   PyTorch BasicVSR SPyNet OpenCV FastAPI NumPy

➜  View the repository

Production-shaped AI systems beyond vision — retrieval pipelines and tool-using agents served behind real APIs.

➜  AI Security Investigator — multi-source incident correlation and triage with structured LLM outputs and an evaluation harness.

Deep Learning  Computer Vision  Video Restoration / Super-Resolution  Generative Models  LLM · RAG · Agentic Systems  Production ML Engineering

M.Sc. Computer Science — Ben-Gurion University of the Negev · in progress

B.Sc. Computer Science — College of Management Academic Studies · 2023–2026 · GPA 92

_{Specialization in Deep Learning · Relevant coursework: Machine Learning · Deep Learning · Computer Vision}

_{Colman Dev Club — collaborative full-stack projects with React, Node.js, Express and Git.}

LinkedIn  ·  GitHub  ·  Email

_{Build. Train. Evaluate. Ship.}