1
1
United Kingdom
4+ years of experience
AI Systems & Product Leader building governance-first AI infrastructure, agentic systems, and safe execution layers. I design and ship high-assurance AI products—combining RAG pipelines, neuro-symbolic reasoning, observability, and control-plane architectures. Previously led digital transformation and AI systems across utilities, fintech, and enterprise platforms. Passionate about building reliable, governed AI agents that operate in real-world environments.

Sovereign Robotics Ops solves the critical gap between AI planning and safe robot execution. When Gemini Robotics 1.5 plans an action like "move to zone B," our governance layer intercepts it BEFORE the robot moves. THE PROBLEM: Autonomous robots powered by AI can make dangerous decisions - moving too fast near humans, ignoring obstacles, or violating safety zones. Current solutions either block AI entirely or trust it blindly. OUR SOLUTION: We sit between the AI brain (Gemini) and robot actuators as the "robot conscience": 1. INTERCEPT - Every AI-planned action passes through our governance API 2. EVALUATE - Real-time policy checks: human proximity, speed limits, geofencing 3. DECIDE - SAFE (execute), SLOW (reduce speed), STOP (halt), or REPLAN (new path) 4. AUDIT - SHA-256 hash chain logs every decision for compliance KEY FEATURES: - <100ms policy evaluation latency - Risk scoring with 0.70 threshold enforcement - Tamper-proof audit trail for regulatory compliance - ISO 42001 & EU AI Act ready compliance reports - Mock fallback when Gemini unavailable - always works TECHNICAL ARCHITECTURE: - Frontend: Next.js + Tailwind (Vercel) - Backend: FastAPI + PostgreSQL (Fly.io) - Integration: Gemini Robotics 1.5 API with mock fallback - Simulator: Gazebo/Isaac Sim ready via WebSocket TRACK 1 ALIGNMENT: Our governance layer evaluates and controls robot actions, processes sensor data for human detection, reacts to environmental changes in real-time, and enables safe sim-to-real transfer through consistent policy enforcement. We don't replace Gemini - we make it safe for production deployment.
15 Feb 2026