Anuj. Bareilly. One laptop. 21 MCP servers, 11 free LLM providers, ₹0 inference. This is M4ST — an autonomous agent system that ships while others plan.
No investor deck. No hiring plan. No cloud credits. Just a laptop, a free-tier routing chain across 11 providers, and a system that hasn't cost a rupee to run since day one.
While everyone else provisioned GPU clusters, I stitched together a failover chain that keeps inference at ₹0 — and it's been running for 18 months straight without a single paid API call.
CEH-track. Self-taught. Jugaad-first. Shipped beats planned.
Multi-agent systems using CrewAI, LangGraph & AutoGen. Autonomous, self-correcting, always on.
21 custom protocol servers. Databases, browsers, OSINT — all talking to models directly.
11 providers in a failover chain. ₹0 inference. 18 months running. No compromises.
Quantized models. Consumer GPU. No cloud dependency. Full privacy.
12ms ChromaDB cache. 40% fewer calls. Every question answered once.
CEH-track. Recon pipelines. Real infrastructure audits with real results.
Four systems. All running. All generating real output.
5 agents · MCP orchestration · semantic cache · zero-cost routing. The full stack.
Agentic coding workspace. Wire into GitHub & Vercel. Build → merge → deploy.
OSINT framework. Shodan + AI. Automated recon and analysis.
12ms vector cache. ChromaDB. No question costs twice.
11 providers. ₹0 inference. The bet paid off.
1 server became 21. Everything connected.
M4STCLAW v3. First multi-agent system shipped.
SEPCC + 9Router. The build never stops.
Open for freelance AI/automation, security consulting, and systems that actually ship.