Faizan M.

Right now, 6 AI agents I built are running 50+ tasks a day for one client with nobody watching. That’s my job: AI agents and AI automation on n8n, Hermes and GoHighLevel, with Claude Code inside them. I built the dialer pipeline that reviewed 118,084 sales calls and put 401 qualified leads into a CRM at 98.3% success. I took three dead n8n workflows to 100% success and cut the client's AI cost per article by about 80%. Top Rated, 100% Job Success, 14 Upwork contracts, six of them multi-agent deployments. Every system on this profile is one I designed and delivered, and most are still running. Almost every project that lands on my desk starts the same way. Someone got an AI demo working, everyone got excited, then real data showed up with real volume and edge cases nobody planned for, and it quietly stalled. That gap between "it worked once" and "it runs my operation" is where I work. WHAT I BUILD - Workflow automation in n8n, Make and Zapier: multi-step logic, API integrations and CRM syncing that hold up at scale, with error handling on every node. - Agent teams on Hermes, Grok Bot and OpenClaw: research, outreach, lead scoring and back-office work, with scoped permissions, routines that run on a schedule, and human approval where it matters. - GoHighLevel automation: follow-up by voice, SMS and email, pipeline stages that update themselves, cold leads re-scored and re-engaged, with Claude reading the replies. - AI integration inside your existing stack: Claude API and OpenAI API with cost tracking, OpenRouter fallback routing, FastAPI backends. PROOF FROM PRODUCTION - AnswerManiac. I built 6 specialized agents on one VPS with OpenRouter routing, FAISS retrieval, and a Mission Control dashboard so the owner watches everything and touches nothing. - GoHighLevel follow-up for a real estate platform, built on Hermes. I built two agents: one runs outreach by VAPI voice, Twilio SMS and email and moves pipeline stages, the other scores cold leads and re-engages them. Operator dashboard on Supabase. - Government bid pipeline. I built the daily system that scores 20-40 solicitations a week, prices them against 5 live distributor feeds and packages them for signature. 245 tests, 10 integrations, live runs with zero errors. - Buying-signal engine on n8n. I built the scanner that watches 600 companies weekly, verifies every source in four layers, and sends outreach capped at 10 emails a day. 98.7% conformance to the client's brief. HOW I WORK 1. Short call, then a written plan. You tell me what is breaking. I tell you if it is worth building, and if not, I say so before you pay for anything. 2. First working piece, fast. One part of the flow running end to end within days so you can click it and react. Then progress every few days, not a month of silence. 3. Harden it, then hand it over. Error handling, retries, logging, alerts, docs, a dashboard, and a support window after launch. The part most demos skip. If you want a quick throwaway demo, I am probably not your guy. If you want a system still quietly doing its job six months from now, we should talk. Stack: n8n, Make, Zapier, Hermes, Grok Bot, OpenClaw, Python, FastAPI, Claude API, Claude Code, OpenAI API, OpenRouter, GoHighLevel, Supabase, PostgreSQL, VAPI, Twilio, Deepgram, Whisper, FAISS. Send me a short note about what you are trying to automate. Two lines is enough. I will come back with how I would build it, what it will take, and whether it is worth building at all.