I treated patients and ran hospitals in Hyderabad for twenty years. Since February 2026 I design, build and operate production multi-agent AI, with the discipline a clinic needs: written protocols, safety gates, audits and measured results.
A 15-agent WhatsApp and Telegram platform that captures, qualifies, routes and follows up enquiries, with the owner approving anything sensitive. Live since June 2026.
My self-hosted, governed multi-agent AI operating system: 9 agents for orchestration, research, job search, education, SRE triage and records, on a 2012 Mac mini with 8 GB of RAM.
Autonomy levels, risk classes, approval tiers, kill switches and audit templates, mapped to the OWASP Agentic Top 10, NIST AI RMF and ISO/IEC 42001, with a healthcare overlay for India and a small policy checker.
How an agent fleet's heartbeat tokens fell 91%: 18 cost methods, a measured before and after, and an audit checklist anyone can run on their own fleet.
An administrative-only assistant for small dental clinics: consent-first booking, reminders, recalls and hand-off to a human for anything clinical. It never diagnoses, prescribes or markets.
Each demo runs the same rules as its repository, on invented data. Nothing is sent anywhere.
Pick an autonomy level and an action; see the decision and the audit event it logs.
Approve agent follow-ups, watch quiet hours and human takeover in a single-writer log.
Edit an agent fleet and see what longer intervals and a 0-token sentinel save.
A test-gated release, a memory governor on 8 GB and an owner-only kill switch.
I direct AI coding agents through specs and task cards with acceptance criteria. I own the architecture, review, testing, release and operations.
Autonomy is earned. Anything touching patients, money, customers or data needs a human approval and leaves an audit trail.
Measure first. Deterministic checks before model calls, right-sized models, cache-friendly prompts.
Remote roles, worked from India, in healthcare and dental AI, agentic systems (forward-deployed, solutions) and AI governance. I am also open to advising clinical AI teams on product validation, patient safety and India go-to-market. Flexible hours for UK, EU and US overlap.
Résumé on request: email me with the role and I'll send the version that fits it.