I'm a second-year Computer Science student at the University of York, building software that works in production — not just on localhost. Before my first university lecture, I had already shipped a production API handling 1,000+ daily requests, and gone on to publish research at NeurIPS.
My engineering work spans three intersecting areas:
- Full-stack product engineering — architecting complete systems end-to-end: backend services, web and mobile clients, and the infrastructure that ties them together.
- Applied AI/ML research — quantum-inspired model compression techniques for convolutional neural networks, peer-recognized at NeurIPS 2024.
- Independent ownership — solo-architecting
Discern-Match, a full-stack TypeScript monorepo spanning mobile, web, and admin tooling, with a product engineer's eye for the details that make software feel finished.
I care about the nuances that separate a prototype from a product: error rates, deployment consistency, and the architectural decisions that hold up under real usage.
Open To
- Software Engineering internships — 2026
- Machine Learning Engineering internships — 2026
- UK-based · open to remote and relocation
Languages
Frontend
Backend & Databases
Cloud, DevOps & Tooling
| Domain | Proficiency | Details |
|---|---|---|
| Model Compression | Advanced | Quantum-inspired pruning and tensor decomposition applied to CNN architectures (GoogLeNet, DenseNet, ResNet-18); published at NeurIPS 2024 |
| Computer Vision | Intermediate | CNN image classification on CIFAR-10 using Keras, with data augmentation and performance visualization |
| Applied Research | Advanced | Peer-recognized publication exploring efficiency/accuracy tradeoffs in deep learning inference |
| LLM Integration | Intermediate | Product-level integration of the Anthropic API within the Discern-Match platform |
Discern-Match — AI-powered matchmaking platform
A full-stack TypeScript monorepo solo-architected end-to-end — mobile, web, admin tooling, and backend all built independently.
| Stack | NestJS · Next.js · Expo / React Native · Turborepo · TypeScript · AWS S3 · RevenueCat · Anthropic API |
| Scale | Full-stack monorepo spanning three client surfaces (web, mobile, admin) and one backend service |
| Performance | Turborepo pipeline caching across shared, type-safe contracts between apps |
| Security | AWS S3-backed asset storage with RevenueCat-managed payment entitlements |
| Impact | Architecture, backend, mobile, and admin tooling — built entirely by one engineer |
| Repository | github.com/edwardmagongo/Discern-Match |
Currently in active, ongoing development — a custom design system underpins all three client surfaces, with a single NestJS backend serving web, mobile, and admin consumers.
QIANets — Quantum-inspired AI model compression research
Research initiative applying quantum-inspired pruning and tensor decomposition to convolutional neural networks, recognized at NeurIPS 2024.
| Stack | Python · PyTorch · JavaScript (research site) |
| Scale | Applied across three CNN backbones: GoogLeNet, DenseNet, ResNet-18 |
| Performance | Measurable inference latency reduction while preserving classification accuracy |
| Security | N/A — academic research artifact, no production security surface |
| Impact | Published at NeurIPS (October 2024) as sole author |
| Repository | github.com/edwardmagongo/QIANets-Website |
The research site documents the methodology and results behind the underlying compression technique, making the work accessible beyond the paper itself.
Medvice — Dockerized healthcare REST API
Architected a Dockerized Flask/SQLAlchemy REST API from scratch for Medvice's symptom-checker platform, as the sole backend contributor on a 4-person cross-border remote team.
| Stack | Flask · SQLAlchemy · SQLite · Docker · Flask-Migrate |
| Scale | Structured medical data API (symptoms, age, gender) powering Medvice's healthcare platform |
| Performance | Standardised, containerised deployment consistent across dev, staging, and production |
| Security | Validation logic enforced on all raw medical input, guaranteeing structured, accurate JSON output |
| Impact | Cut database errors by 90%; shipped independently within a 3-month internship |
| Repository | Private — company-owned codebase |
Work typically handled by mid-level engineers, delivered end-to-end during a 3-month internship — including data modelling, API design, and migration management via Flask-Migrate.
Quantum Compression — Efficient CNN compression techniques
The core research engine behind the NeurIPS-published paper, implementing the quantum-inspired pruning and tensor decomposition methods applied in QIANets.
| Stack | Python · PyTorch · NumPy |
| Scale | Compression techniques evaluated across multiple CNN architectures |
| Performance | Reduced inference latency while preserving model accuracy |
| Security | N/A — academic research artifact |
| Impact | Core technique behind "Quantum-Inspired Techniques for Efficient Model Compression in Convolutional Neural Networks", published at NeurIPS 2024 |
| Repository | github.com/edwardmagongo/Quantum-Inspired-Model-Compression |
Explores the tradeoff space between model size, inference latency, and accuracy — the foundation for the NeurIPS submission.
June 2024 – August 2024 (3 months) · Netherlands (Remote)
Backend contributor on a 4-person cross-border remote team, architecting a Dockerized Flask/SQLAlchemy REST API from scratch for Medvice's symptom-checker healthcare platform.
- Designed data models and API layer for structured medical data (symptoms, age, gender)
- Containerised the application stack with Docker and docker-compose for consistent local development
- Implemented Flask-Migrate database migrations to manage schema changes reliably across the project
- Collaborated on validation logic to ensure structured, accurate JSON output from raw medical input
- Cut database errors by 90% and standardised deployment across dev, staging, and production
Flask SQLAlchemy Docker SQLite REST API Design Flask-Migrate
June 2024 – July 2024 (2 months) · Dar es Salaam, Tanzania
Delivered end-to-end Flutter/Dart UI features within a 5-engineer agile team, extending cross-platform product coverage.
- Improved mobile app load times by 15% via image caching, lazy loading, and state management
- Delivered features to 50+ beta testers, improving onboarding experience
- Participated in full agile development sprints from feature scoping through QA testing
- Built responsive web and mobile features with Firebase integration for consistent cross-platform UX
Flutter Dart Firebase Agile State Management Cross-Platform Development
| Recognition | Details |
|---|---|
| NeurIPS Publication (Oct 2024) | "Quantum-Inspired Techniques for Efficient Model Compression in Convolutional Neural Networks" |
| Production Impact | Cut database errors by 90% as sole backend contributor at Medvice |
| Early Production Experience | Shipped a production API handling 1,000+ daily requests before starting university |
| Project Delivery | 6 projects shipped end-to-end, spanning research, full-stack, and mobile domains |
learning:
- Distributed systems and scalable backend architecture
- Advanced LLM application patterns with the Anthropic API
building:
- Discern-Match — full-stack TypeScript monorepo (NestJS, Expo, Next.js)
exploring:
- New areas across AI/ML research and full-stack systems design
open_to:
- Software Engineering internships, 2026
- Machine Learning Engineering internships, 2026
- UK-based, open to remote and relocation
"Designing intelligent systems by focusing on the unique nuances which bring products to life."