A privacy-first, fully offline neural machine translation application supporting 200+ languages. Built on Meta's NLLB-200 model, this tool runs entirely in your browser without requiring server infrastructure or internet connectivity after initial model download.

Pixel art editor with AI generation. Experiment testing LLMs for structured pixel output via WebGPU in the browser. Two models: OpenAI API (GPT-5.2) and local Ministral 3B via Transformers.js. Spoiler: small local models don't work great for this, but it was fun to build. Not for production use.

A client-side Agent Observability platform. Monitor agent traces, track token costs, and detect semantic drift in real-time. Features 100% local inference via WebGPU and chaos engineering tools.

Privacy-first AI knowledge workspace that runs fully client-side with local PDF ingestion, IndexedDB persistence, worker-based search, and on-device summarization and Q&A.

Browser-native AI demo: chat to generate Mermaid diagrams rendered on an Excalidraw canvas using WebMCP, Gemini, and on-device AI

Local batch image analysis in the browser using Ministral 3B

Browser-based audio similarity search — no server, no uploads, runs entirely in your browser.

Privacy-first AI-powered document search. Upload docs, search semantically with Transformers.js - all client-side, no data leaves your browser.

Intelligent RSS feed aggregator with automated fetching, MongoDB storage, and REST API for real-time content processing.

Browser-based image classification example application using OpenAI CLIP and Transformers.js. Upload images, run zero-shot classification with custom labels directly in the browser, and store images plus results in S3-compatible Backblaze B2 cloud object storage using pre-signed uploads.

Browser-first background removal SDK with one-liner and client APIs, powered by WebGPU/WASM in the browser.

👁️ Client-Side Computer Vision: Turning 2D images into 3D sculptures in real-time. ⚡ Powered by WebAssembly (WASM), ONNX & Transformers.js (Zero-Backend).

LucidLLM is a browser-based LLM chat application supporting local inference via WebGPU/WASM. It features multilingual support (i18n), session management, OPFS model storage, Google Drive backup, and real-time token speed monitoring. Built with vanilla JavaScript and Tailwind CSS.

8 production-grade AI/ML projects: fine-tuned LLMs (DistilBERT, LLaMA, Code Llama, RoBERTa), RAG systems, NLP pipelines — each with FastAPI backend and live inference endpoints.

🚀 A private, local-first ReCAPTCHA solver using Transformers.js and Whisper ASR. Feature-rich with intelligent automation and Chrome Side Panel integration.

Fullstack AI-powered job matching platform. Scrapes Polish job portals, ranks offers using vector embeddings and cross-encoder re-ranking. Turborepo monorepo.

One command. Zero cloud. Your docs stay private.

Real-time video object detection monitoring demo using Transformers.js and Backblaze B2 Cloud Storage. Runs YOLOv9 fully in the browser with automatic capture intervals, no server or GPU required. Detected snapshots and metadata are stored in low cost, scalable B2 storage for analysis and retention.

🚀 Build modern API applications effortlessly with this lightweight, PSR-compliant PHP framework featuring routing, ORM, and secure authentication.

Your music player deserves a glow up.