TypeScript • React • Next.js • Python • WebGPU • PyTorch • transformers.js

Frontend & ML Engineer focusing on interactive AI experiences, multimodal interfaces, and real-time ML inference in the browser.

I combine advanced frontend engineering with practical machine learning to build applications where AI feels instant, visual, and intuitive — from WebGPU-powered inference to multimodal creative tools.

• Browser ML Inference: WebGPU, WebAssembly, transformers.js
• AI-Powered Interfaces: Real-time UI, generative design tools
• Multimodal UX: Text, image, audio, camera input, drawing
• Frontend Architecture: Next.js apps with streaming, RSC, and WebSockets

Core:
  - TypeScript • React • Next.js 14 (App Router)
  - TailwindCSS • Framer Motion
  - Zustand • TanStack Query

Graphics & Visualization:
  - Canvas API
  - WebGL • Three.js
  - WebGPU (experimental)

Real-Time UX:
  - WebSockets
  - Server-Sent Events
  - Streaming UI (React Server Components)

Browser ML:
  - transformers.js (LLMs & vision models)
  - ONNX Runtime Web
  - WebGPU / WASM inference

Python ML:
  - PyTorch • Diffusers
  - OpenCV • torchvision
  - HuggingFace pipelines

AI Applications:
  - Embeddings • Vector search
  - RAG • LLM agents
  - Multimodal processing (text, image, audio)

Backend:
  - Python • FastAPI
  - Node.js • Express

Databases:
  - PostgreSQL • Redis
  - Vectordbs: pgvector, ChromaDB

DevOps:
  - Docker • GitHub Actions
  - Metrics & monitoring basics

• Building client-side ML applications (WebGPU / transformers.js)
• Designing real-time multimodal UI
• Dynamic visual interfaces powered by embeddings/LLMs
• Browser-first inference, offline/edge ML workflows

• Next.js apps with responsive, high-performance UI
• Real-time interactions (WebSocket/SSE streaming)
• Canvas/WebGL for interactive graphics
• UI for ML tasks: visualization, prompting, data inspection

• FastAPI APIs for ML inference
• Python pipelines for preprocessing and embeddings
• Vector search for semantic applications

Technical Stack:

Frontend:
  - Next.js 14 • TypeScript
  - Canvas & Three.js for visual tools
  - WebSocket-based streaming
  - Framer Motion animations

Browser ML:
  - transformers.js for on-device LLM/Vision
  - WebGPU acceleration
  - ONNX Runtime Web for fallback

Backend:
  - FastAPI • PyTorch
  - Custom accelerated diffusion pipeline
  - Redis cache • PostgreSQL storage

Key Features:

• ⚡ Instant AI feedback with browser-based inference
• 🎨 Drawing canvas + text + image → combined AI output
• 👁 Real-time visual previews with progressive rendering
• 🔄 Hybrid engine: browser ML + Python API for heavy tasks
• 👥 Collaborative mode via WebSockets

• WebGPU acceleration for LLM inference
• Add on-device multimodal pipelines (image/text/audio)
• WASM runtime for diffusion-lite model

• Fully-streaming UI for AI apps (RSC)
• Advanced visualization tools (Canvas + WebGL)
• AI-driven UI personalization

• RAG chat with visual memory
• Smart prompting helpers inside UI
• Audio-to-visual creative mode

• CI/CD + automated preview builds
• Load testing for streaming features
• Vector database improvements