Software engineer building
frontier AI and low-latency systems

About

I'm a second-year Computer Science student at King's College London, drawn to frontier AI and the systems work that makes it actually usable. I've shipped voice AI and LLM-driven products at hackathons, and I'm now going lower-level, building a C++ matching engine to understand latency from the metal up. I'd rather ship one finished, correct thing than ten clever demos.

Recognition

• 3rd of 74 at ElevenHacks (AWS Kiro) Apr 2026

  Built Hearsay solo. Top 4% of entries. $2,330 prize ($2k cash plus ElevenLabs Scale).

Selected work

• Hearsay

  3rd / 74

  Can an AI catch you bluffing out loud, in real time?

  Voice-bluff party game. Real-time TTS/STT with ElevenLabs and a hybrid LLM/deterministic game brain (Gemini Flash) that judges spoken claims.

  • Next.js
  • TypeScript
  • ElevenLabs
  • Gemini
  Live ↗ Repo ↗

• Vouch

  in progress

  How do two people close a private deal without one of them getting scammed?

  Voice-confirmed payments for freelancers and private sales: the buyer's money is held until the deal is confirmed by voice, then released. A deep Stripe integration (Connect, Payment Intents, Transfers, Issuing, webhooks) with Vera, an agent on ElevenLabs ConvAI. Started at ElevenHacks and I'm building it out over the summer.

  • Next.js
  • Stripe
  • ElevenLabs ConvAI
  Live ↗ Repo ↗

• C++23 matching engine

  planned

  How fast can a limit-order book match, and how do you prove it?

  Provisional scope, design doc still to come: a C++23 price-time-priority matching engine with a lock-free ring buffer, intrusive order book, and deterministic LOBSTER replay. The details here are a placeholder until I work through the design.

  • C++23
  • lock-free
  • low-latency