I find the repetitive workflows in a business that AI can take over, then build and deploy the complete system around the tools the business already uses: CRMs, databases, email, calendars, documents, websites and internal APIs.
I don't ship demos. I ship AI systems that are reliable, observable, secure and maintainable in production, with human approval wherever it matters.
$30,000+ earned on Upwork · 100% Job Success · Top Rated
311+ GitHub stars · 109 public repos · 3 npm packages
10,000+ documents/day processed in production · 99.2% less processing timeWhat I build: AI agents and multi-agent systems · MCP servers and integrations · LangGraph workflows · RAG and knowledge systems · voice AI agents · business automation · full-stack AI apps (FastAPI + Next.js)
| Project | What it does |
|---|---|
| YouTube Insights MCP | 🆕 Lets Claude analyze any YouTube channel and read your own YouTube Analytics: retention, traffic sources, subscribers per video. Google OAuth 2.1, MCP spec 2026-07-28 |
| WordPress MCP Server | 190+ tools for AI-managed WordPress: posts, pages, media, users, WooCommerce, themes, plugins, SEO, backups and files |
| Context Engine MCP | Code intelligence: semantic search, architecture maps, call graphs, context retrieval and multi-file edits |
| QuickMCP | Production MCP framework with OAuth2, rate limiting, health checks and metrics |
| FLUX Protocol | Universal connectivity layer between AI systems and external tools and APIs |
| code-context-mcp | Smart context and semantic search for AI coding workflows |
| Project | What it does |
|---|---|
| SajiCode | 17-agent autonomous engineering CLI (1 PM, 6 leads, 10 sub-agents) that plans, codes, reviews and ships software |
| MediVoice AI | Real-time voice AI for medical practices: inbound calls, bookings, prescription refills. FastAPI, Pipecat, LangGraph, Twilio, Deepgram |
| WorkFlow AI | Turns spoken workflow ideas into n8n, Zapier or Make.com automations |
| n8nCopilot | AI agent that generates n8n workflows, with structured memory |
| LangGraph Agents Template | Production-ready LangGraph starter with multiple agent patterns and streaming |
| Code to Cash | Multi-agent system that turns product ideas into shipped software |
🌐 More AI apps and tools
| Project | What it does |
|---|---|
| ContextPilot | Context retrieval that reduces hallucinations in AI coding |
| Browser Agent | Vision-first browser automation agent |
| WhatsApp Bot | AI WhatsApp chatbot with admin dashboard and analytics |
| Personal WhatsApp Assistant | WhatsApp + LangChain personal assistant |
| AI Design Assistant | Gemini inside Adobe Illustrator for design analysis and image generation |
| OpenAI Assistant API with UI | Full UI for the Assistants API with threads and streaming |
| eBook Architect | Turns an idea into a complete, structured eBook |
| Upcraft | AI tool that improves Upwork proposals and profiles |
| Apps Generator | Generates full-stack apps from structured specs |
1. Understand the workflow → what is slow, manual or error-prone
2. Design the agent architecture → agents, tools, memory, permissions
3. Build → backend, MCP connections, dashboard, approval flow
4. Connect → APIs, databases, documents, business systems
5. Test on real tasks → not demos
6. Deploy, monitor, improveSecurity I build in by default: OAuth 2.1 · API keys · role-based access · audit logs · rate limits · human-in-the-loop approvals
LangChain: Deep Agents · LangGraph Essentials · LangChain Essentials DeepLearning.AI: Generative AI for Everyone · Building Systems with the ChatGPT API · Prompt Engineering Google: UX Design Certificate · ORCID 0009-0008-1956-0514




