Professional Experience
Product Engineer | Burp AT
PortSwigger
Java (Quarkus)Python (Strands)AWSKubernetesPostgreSQLLLM Agents
- Reduced user approval friction without compromising safety by driving evals-led, end-to-end delivery of an LLM action classifier that auto-approves safe agent actions.
- Own product features end to end, from data-driven hypotheses through implementation, outcome measurement and iteration.
- Steer model, prompt and tooling decisions by building agent evals that measure task completion performance.
Software Engineer (Full Stack)
Cambridge Kinetics
C# .NETTypeScriptAzureMongoDBVueGraphQL
- Reduced customer onboarding configuration from hours to seconds by leading end-to-end delivery of LLM-powered workflow automation, owning requirements scoping, system architecture and team coordination.
- Aggregated customer data from multiple external systems by architecting and implementing a data synchronisation service, processing thousands of daily updates across heterogeneous schemas with minimised API overhead.
- Improved legacy REST API performance by implementing type-safe bulk operations, reducing 100 round-trips into a single call and supporting operations on 10k+ records.
- Cut MongoDB Atlas costs by ~1000£/month by profiling query patterns, identifying excessive search memory consumption from read-heavy workloads, and migrating to dedicated Atlas Search nodes.