B.Sc. Computer Science student at Ariel University (expected Feb 2027). I build end-to-end systems - deep-learning pipelines for physiological time-series, and production web apps that people actually use.
Currently doing AI and product development at IDF Ground Forces HQ, alongside an AI engineering internship at Philosoft.
| Project | What it is | Stack |
|---|---|---|
| SentinelFetal | Real-time fetal-distress detection from CTG signals: 4-stage inference pipeline, <50ms latency, 24 concurrent patients, clinician-facing simulation UI. Current model is a PatchTST foundation model with masked pre-training - AUC 0.839 / Sensitivity 0.818 / Specificity 0.773 on a held-out CTU-UHB test set (model repo private). 1st place, AI For Life Hackathon 2026. | Python, PyTorch, FastAPI, WebSocket, React, TypeScript |
| vscode-rtl-ai-agents | VS Code extension that fixes Hebrew/Arabic RTL rendering in Claude Code, Copilot Chat, Codex and Gemini Code Assist. Webview JS injection with a MutationObserver for streaming responses, plus backup, restore and dry-run modes. | JavaScript, VS Code API, CSS |
| Visual Catan | Settlers of Catan engine: immutable game state, atomic commands, hex graph of 54 vertices and 72 edges, modular rule layer. | TypeScript, React 19, Vite |
| envy-free-room-allocation | Envy-free rent division using maximum-weight matching and longest path in the envy graph. | Python |
| algo-ia | Interactive visualizer for DFS, BFS, shortest paths and cycle detection. | TypeScript, React |
- Languages: Python, TypeScript/JavaScript, Java, C/C++, SQL
- ML: PyTorch, scikit-learn, time-series transformers, LLM/VLM APIs
- Web: React, Next.js, FastAPI, PostgreSQL, Tailwind
- Tools: Git, Docker, pytest, Vitest, Playwright, Azure ML, Vercel