Nathan Wang nwang783

Hi, I’m Nathan Wang

I build a lot. I write about AI.

AI-agent-first e-commerce platform designed for autonomous agents to operate full online stores end-to-end.

Architected API-first store operations spanning agent registration, product lifecycle management, checkout, and analytics
Integrated Stripe Connect for seller onboarding and payouts, plus Printful for on-demand physical fulfillment
Built webhook-driven order, payout, and review pipelines and deployed the system in production on Google Cloud and Firebase
Designed the platform to support both autonomous agent operators and human buyers with a production-ready UX

Repo
Autonomous cybersecurity agent for vulnerability discovery, recon orchestration, and structured security analysis.

Designed a multi-phase agent architecture with deliberate workflow control, structured context management, and long-running execution semantics
Built modular orchestration for recon, testing, and sandboxed tool execution to support production-style security workflows
Emphasized system design quality, observability, and maintainability for autonomous behavior that can run beyond toy demos

Repo
Agentic reverse-engineering system for binary triage, analysis planning, and reproducible binary research workflows.

Built LLM-powered workflows that generate, execute, and adapt structured reverse-engineering plans against real binaries
Integrated Ghidra via MCP alongside sandboxed exploit-testing flows to connect model reasoning with practical tooling
Produced structured findings and reproducible analysis outputs suitable for iterative research and debugging

Repo
Real-time voice tutoring platform combining production web UX with low-latency, multimodal tutoring sessions.

Integrated Daily WebRTC, Deepgram STT, OpenAI LLMs, and Google TTS into a cohesive real-time tutoring stack
Added session continuity, PDF-informed tutoring, and interactive study tools to support sustained learning workflows
Improved reliability, product polish, and live-session UX for production-oriented conversational learning experiences

Live Demo
Award-winning GenAI concept for U.S. asylum/immigration workflows.

Built demos for RAG-based case support, summarization, and custom prioritization metrics
Created and used a synthetic dataset of 71 generated asylum cases for the prototype
Winner of UVA McIntire’s Gen AI for the Common Good Challenge
Coverage: Competition Article

Languages: Python, TypeScript/JavaScript, C
Frameworks & Platforms: React, FastAPI, Firebase, AWS, Google Cloud
AI Systems: Agentic workflows, RAG pipelines, real-time voice agents, model/tool orchestration
Infra Integrations: Stripe, webhooks, API design, production deployment

If you’re building in agentic AI, applied ML, or AI product infrastructure, I’d love to connect.