Computer Science at Cambridge · looking for software engineering internships, summer 2027
View my workI'm Samuel, a second-year Computer Scientist at Churchill College, Cambridge. I like building complete products: a real-time voice app with a React front end and a FastAPI backend, an ML pipeline served through a Django web app, and a retrieval tool over Cambridge past papers.
Whatever I'm building, I care about knowing it works: tests, honest evaluation and real data. On the side I enjoy low-level C++, like an order book checked against a full NASDAQ trading day. Away from the desk: competitive programming, chess and Churchill football.
Python
Java
C++
React
Django
pandas
scikit-learn
BashAlso: XGBoost · LLM APIs & RAG
Screenshots and recordings from the real projects.
Built in a day at Google's Student AI Hackathon with a React front end and a FastAPI backend. Describe an emergency out loud, and seconds later it lands on a responder's 3D globe as a structured, urgency-ranked incident.
Do recent-form stats beat skill rating at predicting Counter-Strike matches? A leakage-safe pipeline tested on 5,000 real matches says no. But isotonic calibration more than halved the miscalibration (0.072 → 0.032), which is what matters when you use the probability.
A limit order book replayed against a full day of real NASDAQ market data and checked message by message against NASDAQ's own executions. 51 ns median per message, 36M messages a second, zero heap allocations.
Ask about a topic and it finds the Cambridge past-paper questions that actually test it, with examiners' notes on where students went wrong. A multithreaded vision pipeline reads the PDFs with the maths intact, and parallelising it cut processing time by 75%.
A day in London building alongside Google engineers. My team's voice-first crisis reporting app was pitched to a panel of senior engineers. See the project →
A week-long, invite-only algorithmic game theory contest where 38 teams write bots that attack and defend. I built a practice arena that rebuilt rival teams' strategies from match logs, so we could test against realistic opponents instead of guessing.
Pitched an idea built on Visa's APIs: an AI model inside banking apps that shows, in real time, how a purchase would move your credit score.
Looked after the Windows Server setup that lets branches share data securely, and stood up a Squid proxy on Ubuntu Server. Tuning its caching and security rules made cached content load 60% faster and cut external bandwidth.
Algorithms I & II, Operating Systems, OOP (Java), Foundations of CS (OCaml), Databases, Software & Security Engineering, Machine Learning.
Mathematics, Further Mathematics, Physics, Chemistry. UKMT Senior Maths Challenge Gold.
Hiring interns for summer 2027, or want to dig into one of these projects? Email is the quickest way to reach me.
