I build AI automations and custom chat agents on top of 5 years of real support-floor experience - so the systems I build fit how support work actually gets done.
WHAT I BUILD NOW:
- Custom chat agents (support, receptionist, booking, lead-capture) designed from a business's own content and brand voice, with clean human hand-off
- AI inbox and ticket triage: classify, prioritize, escalate, draft approval-first replies (3.19s average per message across logged runs)
- Lead-speed automation: a self-hosted n8n + Claude pipeline that takes a form submission to a scored, routed reply in 7.4 seconds, verified across four logged executions
- Workflow automations on self-hosted n8n connecting forms, sheets, CRMs, email and calendars, with webhook and API integrations
THE FOUNDATION UNDER IT:
- 5 years of BPO support operations at Hinduja Global Solutions (-----------): technical support (Sony), billing support with sales targets (AT&T), and real-time analyst / break master - live queue monitoring, adherence and staffing. I did by hand the problem I now automate.
- Remote Upwork work through the 2010s: virtual assistance plus voice, data and medical transcription
-----------: two years of full-time, self-taught retraining into software and AI engineering - three public agent repos on GitHub carrying committed workflow JSON and logged execution times
HOW I WORK: plain-language deliverables, humans stay in the loop, honest metrics only, self-hosted by default - you own the workflow, the prompts and the data.
PROOF YOU CAN CLICK: -----------/mvallautomations - public builds with case studies. Live demo: handlit.app
STACK: Claude API (agents, MCP), Claude Code, self-hosted n8n, Python, JavaScript/TypeScript, SQL, Ollama, Next.js, Docker, Cloudflare, Playwright, Git
Open to full-time on a flexible schedule, part-time, or per-project. If you have a support inbox, a lead form, or a 'we reply within 48 hours' problem, I can show you in one call how an agent would handle it.
