3rd-year Computing Systems student who builds production-grade software across the full stack — from sub-millisecond TinyML inference on embedded hardware to event-driven cloud architectures on Microsoft Azure. Delivered five end-to-end projects applying industry practices including automated testing, CI/CD pipelines, secrets management, and role-based access control. Seeking a Junior Developer, Cloud Engineer, or IT Support role in the UK.
Languages Python · Java · TypeScript · JavaScript · HTML · CSS
Frameworks & Libraries Flask · SQLAlchemy · Flask-Login · Flask-WTF · ReportLab · WTForms
Cloud Microsoft Azure — Functions · Blob Storage · Cosmos DB · Application Insights · Key Vault
Tools & Testing Git · GitHub · pytest · coverage · SQLite · GitHub Actions
Personal finance web app with secure authentication, transaction tracking, custom budget categories, recurring templates, financial analytics (savings rate, month-on-month comparison, spending alerts), two-phase CSV import, and PDF/CSV export. Includes a full pytest test suite covering auth, transactions, imports, reports, and security. View repository →
Cloud-native multimedia platform built for COM682. Initially architected as a serverless Azure Functions API; after diagnosing a platform-level limitation (Linux Consumption plan failing to start across all available Azure for Students regions), migrated the backend to Flask on Azure App Service with zero API contract changes. REST API backed by Cosmos DB (metadata) and Blob Storage (media), secured with JWT auth, bcrypt, and Key Vault-managed secrets via keyless DefaultAzureCredential access. GitHub Actions CI/CD deploys the API to App Service and the frontend to Blob Storage, with Application Insights for monitoring. View repository →
Full-stack health tracking app with activity logging, nutrition tracking, health metrics, goal management, and analytics. Angular frontend, Flask REST API, MongoDB, and JWT authentication across 9 feature modules. View repository →
Console card game for 2–6 players. Each player is dealt 5 cards, discards at least 2, and aims to get their 3-card hand as close to 15 as possible. Features an AI opponent that brute-forces the optimal card combination, colour and suit bonuses, and a persistent top-10 high score table. View repository →
Real-time ambient lighting classifier that runs on an Arduino. An APDS-9960 sensor feeds a 3,648-byte int8-quantised MLP that classifies lighting into four eye-safety categories (optimal, too dim, too bright, high blue-light) and streams results to a live Processing dashboard. Full pipeline: sensor calibration, data collection, Colab training, TFLite quantisation, on-device inference. View repository →
BSc (Hons) Computing Systems — Ulster University · 3rd Year
meharjot78@gmail.com · LinkedIn: linkedin.com/in/meharjot-singh-dev