While I seek employment this is a great time to get all these projects out of my head and into the 'IRL'!
Recently I've been working on Python School — an OOP proof of concept driven by a local LLM — and Discord_Philosopher — autonomous LLM characters debating in Discord, with Discord_Airflow to analyse the output.
That exploration has grown into something more systematic. I'm now building AMC (Agent Management Console) — a local Docker platform for running LLM workloads and testing them at scale. The questions I'm chasing: what can local models actually do? How reliable and consistent are they across many runs? What's the right model for a given task, and where does it break down? A web portal for accessing test results is in progress.
Join me as I explore technologies through building my solutions!
| Projects | ||
| Joint Account Analyser as PDF | Analyse joint bank account .csv and create a report | R, Python |
| Joint Account Analyser as Web | As above but web-based | TypeScript, React |
| Smart Timelapse Pipeline to YouTube | Daily sunrise-to-sunset timelapse, automatic publish to YouTube | C++, Python, Cron, Rsync |
| Moped API | Google Forms → Django API pipeline for moped fuel data, with service notifications | Python, Django |
| Python School | Terminal simulation of a school — OOP model driven by a local LLM. Generates courses, students, exams and newsletters | Python, Ollama |
| RPS League | Live Rock-Paper-Scissors leaderboard and match history dashboard | TypeScript, React, Fastify, SQLite, Docker |
| Discord_Philosopher | Autonomous LLM-powered characters debating in a Discord server | Python, Ollama, Discord API |
| Discord_Airflow | Text analysis and summarisation of LLM-generated debate content | Python, Apache Airflow, PostgreSQL, Docker |
| LLM Research | ||
| AMC — Agent Management Console | Docker platform for managing Ollama workloads and running systematic LLM tests | Python, Docker, Ollama, SQLite |
| amc-runner | Local GPU inference agent — polls SQS for jobs, runs them via Ollama, returns results to the cloud API | Go, Ollama, AWS SQS |
| amc-watchdog | Raspberry Pi daemon that monitors queue depth and wakes the GPU machine via Wake-on-LAN when jobs are waiting | Python, AWS SQS |
| DevOps | ||
| GitOps | ArgoCD deployments | ArgoCD, Bash |
| Observability | Cluster monitoring | Prometheus, Grafana |
| Infrastructure | ||
| K3s | Kubernetes cluster (3 nodes) | K3s |
| Pi0Cam | Raspberry Pi Zero W with camera | Python |
| Edge-NET | Self-contained WiFi network of Pi nodes | OpenBSD, pf, DHCP, MQTT, MicroPython, Python, C++ |
| Tools | ||
| Repo Registry | Central registry for managing repos, syncing and mapping connections | Bash |
My name is Jack Waddington. My background is in Business & Administration, Visual Communication and Operations Management.
Recently I completed studies in Software Engineering at Hive, Helsinki.
Welcome to connect on LinkedIn!