I build across the stack β from native desktop apps and cross-platform mobile clients to custom machine learning systems. I care about clean architecture, local-first design, and shipping things that work end to end.
|
A from-scratch GPT-style language model (~205M params, PyTorch) that trains on a single consumer GPU β paired with a full PySide6 desktop chat app, BYOK routing, and voice input. |
A local-first, peer-to-peer LAN messenger for desktop and Android β no central server required. Raw TCP/UDP discovery, E2EE direct messages, group chat, and P2P voice/video calls. |
Languages
Frameworks & Libraries
Focus Areas
- π§ Machine Learning: Custom transformer architectures & training pipelines
- π₯οΈ Desktop Apps: Cross-platform tools with Electron & CustomTkinter
- π± Mobile: React Native apps with local-first data
- π Networking: Peer-to-peer & LAN-first architectures β no external servers required
"Bring an idea to life β nothing is impossible πͺ"



