I build production-ready AI agents, MCP servers, RAG systems, and custom AI tools that connect LLMs with real business workflows.
I build AI systems that do more than chat.
My work focuses on the layer between LLMs and real business tools:
LLM → Agent → MCP Server → API / Database / Business Tool → Human Approval → Real ActionI specialize in:
- AI agents that use tools, APIs, memory, and structured workflows
- MCP servers that connect AI assistants to real systems
- RAG systems for business documents, knowledge bases, and search
- Multi-agent systems for automation, coding, research, content, and operations
- Custom AI tools with OAuth, permissions, dashboards, logs, and approval flows
- Full-stack AI apps using Next.js, FastAPI, Node.js, TypeScript, Supabase, PostgreSQL, and Docker
| Area | What I Build |
|---|---|
| 🤖 AI Agents | LangGraph, LangChain, tool-using agents, planner agents, workflow agents |
| 🔧 MCP Servers | Model Context Protocol servers for APIs, databases, WordPress, CRMs, files, and internal tools |
| 🧠 RAG Systems | Hybrid search, vector databases, embeddings, document pipelines, business knowledge bases |
| 🧩 Multi-Agent Systems | Research agents, coding agents, content agents, support agents, automation agents |
| 🔐 Custom AI Tools | OAuth integrations, dashboards, role-based permissions, approval flows, secure tool execution |
| 🌐 Full-Stack AI Apps | Next.js, FastAPI, Node.js, TypeScript, Supabase, PostgreSQL, Docker |
$30,000+ earned on Upwork · 100% Job Success Score · Top Rated
311+ GitHub Stars · 109 public repos · 3 NPM packages published
10,000+ docs/day processed · 99.2% processing time reduction| Project | Description |
|---|---|
| SajiCode | 17-agent autonomous engineering CLI. 1 PM agent, 6 lead agents, 10 sub-agents. Built for planning, coding, reviewing, and shipping software. |
| MediVoice AI | Real-time voice AI for medical practices. Inbound calls, booking, prescription refills, FastAPI, Pipecat, LangGraph, Twilio, and Deepgram. |
| WorkFlow AI | Voice-to-automation system that turns spoken workflow ideas into n8n, Zapier, or Make.com automation structures. |
| n8nCopilot | Claude-style AI agent for n8n workflow generation with structured memory and automation reasoning. |
| LangGraph Agents Template | Production-ready LangGraph template with multiple agent patterns, streaming modes, and JSON-based configuration. |
| Code to Cash | LangGraph multi-agent system that turns product ideas into shipped software workflows. |
| Project | Description |
|---|---|
| WordPress MCP Server | 190+ MCP tools for AI-powered WordPress management: posts, pages, media, users, WooCommerce, themes, plugins, SEO, backups, and file operations. |
| Context Engine MCP | Code intelligence MCP server with semantic search, architecture mapping, call graphs, context retrieval, and multi-file editing support. |
| QuickMCP | MCP server framework with authentication, OAuth2, rate limiting, health checks, metrics, and production-ready structure. |
| FLUX Protocol | Universal AI connectivity layer for connecting AI systems to external tools and APIs. |
| code-context-mcp | Smart context and semantic search MCP server for AI coding workflows. |
| ContextPilot | Context retrieval and code intelligence system for reducing hallucinations in AI development workflows. |
| Project | Description |
|---|---|
| OpenAI Assistant API with UI | Full UI wrapper for OpenAI Assistant API with thread management and streaming. |
| eBook Architect | AI-driven web app that transforms ideas into full structured eBooks. |
| Browser Agent | Multimodal browser automation agent with vision-first navigation and coordinate-based actions. |
| WhatsApp Bot | Full AI-powered WhatsApp chatbot with admin dashboard and analytics. |
| Personal WhatsApp Assistant | WhatsApp + LangChain + GPT personal AI assistant. |
| AI Design Assistant | Gemini AI inside Adobe Illustrator through a CEP extension for design analysis, image generation, and color intelligence. |
| Upcraft | AI tool for Upwork users to improve proposals, profiles, and freelancing workflows. |
| Apps Generator | Coding app generator that creates full-stack applications from structured specs. |
AI / Agents LangChain · LangGraph · LangSmith · MCP Protocol · RAG · Vector DBs
Embeddings · Multi-Agent Orchestration · Tool Use · Function Calling
LLMs OpenAI · Anthropic Claude · Google Gemini · Qwen · Ollama
Backend Python · FastAPI · Node.js · TypeScript · REST APIs · WebSockets · SSE
Frontend Next.js · React · TypeScript · Tailwind CSS · shadcn/ui
Databases PostgreSQL · MongoDB · Supabase · Pinecone · ChromaDB · Redis · SQLite
Infrastructure Docker · GitHub Actions · Vercel · DigitalOcean · AWS
Security OAuth · API Keys · Role-Based Access · Audit Logs · Rate Limits
Human-in-the-loop Approval Workflows
Tooling Claude Code · Cursor · Rust MCP Servers · tree-sitter · Custom CLIs- Deep Agents · LangChain
- LangGraph Essentials · LangChain
- LangChain Essentials · LangChain
- Generative AI for Everyone · DeepLearning.AI
- Build Systems with ChatGPT · DeepLearning.AI
- Prompt Engineering · DeepLearning.AI
- Google UX Design Certificate · Google
1. Understand the workflow
2. Design the agent architecture
3. Define tools, memory, permissions, and MCP connections
4. Build the backend, dashboard, and approval flow
5. Connect APIs, databases, documents, and business systems
6. Test the agent against real tasks
7. Deploy, monitor, and improveI do not just prompt AI.
I build the systems that allow AI to safely work with real tools, real data, and real business operations.
- Custom AI agent development
- MCP server development
- LangGraph and LangChain agent systems
- RAG chatbot and knowledge base systems
- AI social media agents
- AI customer support agents
- AI workflow automation
- WordPress and WooCommerce AI agents
- OAuth-based AI tool integrations
- Full-stack AI SaaS MVPs
- Multi-agent automation systems



