Jack Waddington

Software Engineering Graduate

• I covered topics from Bash, C, C++, Docker, Typescript, React and the full software lifecycle, and how to apply them in fullstack projects. Using solo, self-guided and team study, the course spanned low-level fundamentals all the way to producing coursework with the group.
• Self guided learning forced me to be disciplined and directive with my time and goals, while building, debugging and launching projects
• Working with peers with varied confidence and ability, meant that I was switching between student and mentor roles with a diverse cohort of students
• As part of 42network, the course actively focused on technical primitives (memory allocation, pointers, sockets etc.) , giving me a solid foundation of knowledge to work my way up from low to high level languages like Python
• As the intake of the school grew, I produced handbooks and onboarding manuals to support current and future students to keep up with the changing operational processes in the academy
• Attended many regular events and presentations from leaders across the software community in Helsinki

Operations and IT Manager

I joined this local restaurant when they opened their second location. The popularity of the business scaled overnight, meaning a shift in my role to overseeing operations as they went on to open over 70 locations in five countries. Transitioning to a franchising model led to rapid growth with huge changes in the technology maturity of the business. Our reliance on data became critical to optimise menus and ingredients, promotions, loyalty and underlying business metrics.

• I designed and installed front-and-back of house IT installations and networks across Finland, while defining manuals and processes for international openings
• Partnered with Micros Oracle to migrate and upgrade systems across all locations in line with business requirements
• Integrated and interfaced multiple third parties to provide online ordering (both in-house and partner platforms), kiosks, discounts and promotions
• To support the demand for accurate, consolidated insights across the business, I introduced Looker Studio, automated reporting using Python and R, and advocated for high data quality across all departments, replacing error prone and out of date spreadsheets
• Founded a data team, training two colleagues on SQL, our schemas and building shared business dashboards while implementing agile team methodologies to ensure rapid, quality outcomes
• Knowledge and expertise in the data allowed me to lead conversations to forecast sales and optimise our unit economics through recipe refinements and supply chain adjustments
• I scaled the marketing pipeline to increase throughput, allowing the business to try more ideas in a shorter time by using shared campaign launch processes, data capture and analysis.
• Used this data to support C-suite to navigate tough and unknown decisions during COVID-19
• And plenty of varied tasks across the whole business, from product and subbrand launches, franchising agreements, seasonal changes, designing storefronts, logos and more

With huge attention on expensive, cloud-based frontier models, I built this to allow anybody to trial, compare and evaluate multiple consumer-grade LLMs through a web interface with all inference performed on my local GPU and hardware. To solve the global-to-local leap, I leaned on AWS SQS and integrated a wake-on-lan trigger to save energy costs. It currently supports more than twenty models and tracks key metrics on performance, speed, latency and quality. This became a complex system, which required full CICD using Github Actions and deploying using an IaC configuration on Render. To support debugging this distributed system, I added Sentry exception tracking and logging, and used Mixpanel to understand customer behaviours within the app to help me prioritise feature development.

My current infrastructure includes a

• Static Typescript/React console frontend
• Fastify API server and database also hosted in Render
• A Python daemon running on a low-power Raspberry Pi listening to SQS events, triggering a wake on lan event to my home PC
• A Go server processing user requests on my GPU and pushing LLM results back to the API

I gave myself the challenge of producing a fully-local, secure bank statement processor and analysis tool. By leveraging modern browser APIs, I was able to overcome the stigma of sharing personal data with an unknown or untrusted provider. Based on an existing report I created in R, this guides a user through removing duplicates, categorising and grouping transitions, and provides a clear document showing incomes, outgoings and trends, with easy to follow graphs and charts. One key feature supports navigating multiple incomes in a joint account, making it easy to follow and balance contributions over time.

I run a fleet of Raspberry Pi devices in my house. I've used them for running a timelapse camera from my window with a daily automated scheduled Youtube upload showing the seasons, wildlife, weather and environment change over time. I've collected and I run a number of these devices and I'm currently researching and implementing MQTT to network and connect them. I run many personal projects locally - I started with proxmox, then moved to a kubernetes cluster, managed through Terraform and ArgoCD, monitored using Prometheus and Grafana, and accessible via Tailscale.