Open-source enthusiast
Working on agent tooling, RAG pipelines, and video-translation systems β and breaking my self-hosted server on the side.
- vid_translate β a Tauri desktop app doing live speech translation (JA/ES β EN) with Vosk streaming ASR and CTranslate2, packaged for Linux and Windows via a GitHub Actions release matrix.
- Poking at a FOSS community platform (FastAPI + PostgreSQL + React/Vite) in my spare time.
- Open to new SDE / AI-agent-tooling internship opportunities.
| Project | What it does |
|---|---|
| vid_translate | Tauri desktop app for live JA/ES β EN video translation; Vosk ASR + CTranslate2 (int8-quantized MarianMT), paced-reveal captions, cross-platform CI release pipeline (AppImage/.deb/.msi) |
| RAG-Model | Multimodal document-intelligence RAG pipeline β marker for layout-aware PDF parsing, qwen3-vl for image captioning, bge-large-en-v1.5 embeddings into ChromaDB, served locally with gpt-oss:120b |
| KIIT_WIFI_ISSUE | PowerShell automation for KIIT campus WiFi login/reconnection β cut reconnection time from ~90s to ~4.5s |
| Screen-usage-widget | A small desktop widget for tracking screen-time, built because nothing off-the-shelf did what I wanted |
| Text-Extractor-for-mac-os | Region-select screen OCR widget for macOS |
(full list in my repositories tab)
Self-host a small home lab on Ubuntu Server (CasaOS, Jellyfin, Tailscale, Cloudflare Tunnels), tinker with Fedora/GNOME, and grind DP/binary-search problems in Java and Python when I need a break from agents and pipelines.

