Born out of my frustration to accurate maintain time sheets and from my obsession with tracking all aspects of my life, this simple web app allows a user to: track their work sessions via a stopwatch or time, label each session with customisable metadata, analyse their trends over time to help with designing an optimal personalised work schedule. The long-term vision is to build integrations with health & fitness trackers and correllate them with the measured data, which will allow for even more accurate personalised suggestions.

Based on Backstage by Spotify, this solution was built to provide employees a simple and clear way of managing the company's private cloud infrastructure, leading to a 100x improvement on developer efficiency in some scenarios. This was a truly holistic solution: data modelling and cataloguing all company resources (employees, compute, storage etc.); frontend UI (human users); backend API (machine users) with async messaging able to handle provisioning failures, cancellations etc.; infrastructure provisioning engine via Terraform.

Having a rock solid secrets management solution is key when aiming to build "high integrity" systems i.e. where no single user can obtain privileges to do any serious harm. In my first role at G-Research, I was developing secure data access patterns as well as managing the infrastructure including the whole CI/CD flow of our on-premise secrets management platform using Hashicorp's Vault as the backbone.

Given a start and end point in space (2D or 3D), this neural network can plan collision-free paths by predicting the intermediate points. The model was trained using Model-Agnostic Meta-Learning and resulted in reduced path length and improved collision rate compared to state-of-the-art as per the research in my Master's Thesis.

As an intern (6 months), I contributed to a project during a big migration of load balancer hardware infrastructure. The product detected all components of the legacy infrastructure and automatically created equivalent AWS resources in the new infrastructure as part of a multi-step pipeline.