Building, researching, and shipping.

ARU ARM AI Lab, Cambridge - Apr 2026 – Present

• •Built and optimised a multi-model inference pipeline on a Cortex-M55 platform with a neural accelerator (embedded C/C++, PyTorch), running three AI vision models concurrently in real time within strict latency and memory budgets — validation methodology prepared for IEEE Embedded Systems Letters.
• •Diagnosed complex hardware/software failures (flash addressing, cache coherency, runtime mismatches) down to register level, turning them into a documented debug procedure adopted by the team.

Build Advance Create (self-employed) - Nov 2025 – Present

• •Founding an early-stage SaaS venture: a cloud-connected computer-vision platform analysing CCTV feeds for customer-selected events (fire, theft, injury detection) with automatic emergency-service alerting.
• •Leading product definition, technical architecture (cloud inference pipeline, per-feature detection models), and the subscription monetisation model.

Anglia Ruskin University, Cambridge - Dec 2025 – Present

• •Designed, trained and evaluated CNN models on a custom dataset built from scratch (collection, cleaning, labelling, augmentation) to automate urban waste-stream classification for Cambridge City Council.
• •Benchmarked accuracy, latency and hardware trade-offs for deployment; presented findings to non-specialist stakeholders.