Software

at_c

The alpha C implementation of the atSDK for the at_platform. It is a work in progress and not ready for production use. The atSDK is a set of libraries for building applications on the at_platform, designed to be easy to use while giving users a high level of security and privacy.

What I did

I contributed the beginning of the ed25519 implementation and added test cases for RSA 4096 operability.

EcoCheck (Clean Choice)

Clean Choice is a web app that suggests environmentally friendly alternatives to products you've bought. Generating a receipt creates a QR code that uses the receipt details to show the carbon footprint of your purchase and recommend greener alternatives.

Federated Learning for Adaptive Road Efficiency (FLARE)

Traffic lights that aren't coordinated across a city cause delays. FLARE proposes a decentralized network of IoT devices running a federated learning framework. The devices work in a secure, self-contained network, talking only to each other and the learning algorithm, and adjust signals using real-time traffic data and inflow-outflow equations.

I built a simulation of downtown Victoria in SUMO, explored peer-to-peer network designs in OMNeT++, and documented the work in bi-weekly updates.

Machine learning theory

Problem sheets across the topics below, completed to reinforce my understanding and application of the theory.

• PAC learning
• Consistency model, perceptron, mistake bound, and a PAC learning algorithm for linear separators
• Agnostic learning
• Uniform convergence (finite classes), including Markov's inequality and Hoeffding's inequality
• Uniform convergence (infinite classes): VC dimension
• Rademacher complexity
• Support vector machines, margin bounds and kernels
• Boosting
• Information theory: Shannon entropy and KL-divergence
• Occam bounds and PAC-Bayes
• Compression bounds
• Active learning
• Decision-theoretic online learning
• Prediction with expert advice
• Bandits
• Stochastic and online convex optimization
• Online-to-batch conversion

ATP Tour Analytics 2022

I'm a big tennis fan and wanted to analyze ATP Tour data. For a data science course I analyzed 2022 ATP Tour results and built visualizations that show trends and patterns.