What I Bring
AWS & Cloud
Hands-on AWS projects covering serverless, networking, compute, storage and infrastructure as code.
AI & RAG
Amazon Bedrock, RAG architectures, knowledge bases, document intelligence and LLM integrations.
Automation
AI agents, scheduled research, GitHub automation and automated publishing workflows.
IT Support
Two years of hands-on IT support alongside a decade of customer-facing experience.
Featured Projects
Production Serverless RAG AI Assistant
A production-style serverless AI assistant built on AWS. The application combines Amazon Bedrock Knowledge Bases with Lambda, API Gateway, Cognito authentication, DynamoDB state management and private S3 document storage.
Additional description: The system also integrates booking state with Google Calendar, demonstrating how AI can be connected to a practical business workflow.
View Project DetailsReddit Tech Radar
Automated technology intelligence from Reddit â researched, verified, structured, and published daily. The system uses Reddit as a discovery signal for technology discussions, then applies additional research and verification from primary sources before producing structured daily technology reports.
Additional description: Built with the Hermes agent skill system to automate scheduled research workflows. Uses Reddit discovery followed by verification against official documentation, GitHub repositories, and technical sources to create traceable technology intelligence.
Jason AI Phone Agent
AI-powered voice support system using Vapi for AI voice orchestration, Twilio for UK telephony, and planned AWS backend services for data processing and automation.
Additional description: The Vapi assistant has been successfully tested via browser voice interface with clear, natural conversation. Twilio regulatory bundle submitted and awaiting approval.
Bedrock Document Intelligence
An AI-powered document analysis system using Amazon Bedrock to analyse documents stored in Amazon S3 and persist structured analysis results in DynamoDB.
View Project DetailsRimWorld AI
An AI chatbot and colony advisor for RimWorld, powered by RimMind and OpenRouter. Explores AI integration with RimWorld, documenting capabilities and limitations.
AWS Networking Lab
Hands-on AWS networking work exploring public and private subnets, routing, Internet Gateway and NAT connectivity, security controls and the architecture decisions required to connect resources securely.
Additional description: Built to move beyond memorising AWS networking components and develop practical understanding of how they work together.
View Project DetailsArchitect Floorplan — Hermes + Blender
An experiment in giving an AI agent a structured architectural specification and having it generate a 2D floor plan in Blender — documenting the failures as honestly as the result.
Additional description: The final plan is not the interesting part. The value is in where the agent failed: inventing architecture without constraints, and reporting success on a render that was actually blank.
View Project DetailsJobMatch ML — Explainable Job Matching
A Python and scikit-learn machine-learning project that helps a job seeker prioritise job adverts by estimating how closely each role fits their profile.
What the ML does: TF-IDF converts advert text into numerical features and a Logistic Regression classifier estimates text relevance. The app combines that signal with recognised skill overlap and experience fit to produce an explainable suitability score.
The results highlight matching skills and possible gaps. Example tests compare a strong Junior AWS Cloud Support Engineer match with a weaker Retail Store Manager match. Scores are suitability estimates, not hiring probabilities.
Local LLM Lab — Ollama, Granite & Hermes
A hands-on experiment running the Granite 4 3B language model locally through Ollama, exploring whether a small local LLM could help with AWS Solutions Architect Associate (SAA-C03) practice questions.
What I learned: No model tested was reliable enough to act as a 100% accurate exam answer engine. I also worked through integration problems connecting the local model to Hermes. A Python reporting script eventually provided a practical route to run the local test and capture its output.
The work was then connected to the Hermes automation workflow: scheduled-job results are collected into a daily report and delivered to my private email using AgentMail, helping me see which cron jobs succeeded and which failed.
View Project DetailsGranite — Local AI Voice Assistant
A practical Windows voice assistant built with Python, Faster-Whisper speech recognition, the local Granite 4 3B model through Ollama, and Piper text-to-speech. Granite can save and read notes and to-do items, search file names, and open selected Windows apps or websites with confirmation.
What I learned: Integrating local speech recognition, an LLM and speech playback; debugging real microphone and indentation issues; and documenting speech-recognition limitations such as confusing “buy” and “by”.
AI Command Centre - Local Model Experiment
A browser-based project management dashboard built with Claude Code, Ollama and Qwen 3 8B. Features project management, task checklists, progress tracking, notes and JSON export/import.
What I learned: Testing local AI coding models and verifying that generated files were saved and worked correctly. Project data stays in browser local storage.
Try the Demo Read Project README & DocumentationJase Vision AI - Local Vision Assistant
A browser-based AI vision assistant powered by MiniCPM-V running locally through Ollama. Visitors can explicitly grant webcam access or upload a JPG, PNG or WEBP image and receive a concise scene description from the local model.
Architecture: Python and Flask handle requests, Ollama serves the vision model, and an access-controlled Cloudflare Tunnel provides an invite-only HTTPS entry point. The source is public; the live demo remains restricted to approved testers.
Pooled — Distributed AI Inference Experiment
A hands-on experiment exploring whether AI inference can be distributed across a Windows PC and an iPhone using Pooled. The work investigates browser-based peer connectivity, device participation and shared compute capacity.
Current status: The interface displayed combined allocated capacity, but end-to-end distributed model inference has not yet been verified. The findings page records what worked, what failed and what still needs testing.
DeepSeek Security Auditor — AI-Assisted Code Security Scanner
A Python static-analysis tool that checks source code and configuration for common security risks, including hardcoded credentials, SQL injection patterns, insecure HTTP references, missing security headers and unsafe configuration.
Development story: The initial implementation was generated with DeepSeek. I then used ChatGPT for independent code review, security fixes, regression testing and documentation after early tests exposed missed credential formats and reporting issues.
Verified test run: 98 tests passed, 0 failed on 2 November 2026. This is evidence against the included test cases, not a guarantee of real-world detection accuracy.
More AWS Projects & Labs
Serverless API with CloudFormation
A fully serverless CRUD API deployed entirely with AWS CloudFormation, demonstrating Lambda, DynamoDB, API Gateway, and IAM roles using infrastructure-as-code.
View Project DetailsPrivate Serverless VPC
A secure AWS serverless application running inside a private network where Lambda functions communicate with DynamoDB without public internet exposure.
View Project DetailsRetrieval-Augmented Generation (RAG)
Two RAG implementations: enterprise-grade with Pinecone vector database and a basic implementation using Gemini API with custom knowledge base.
View Project DetailsECS Fargate Deployment
Containerized Python FastAPI microservice deployed on AWS Elastic Container Service using Fargate launch type with ECR, IAM, VPC networking.
View Project DetailsMulti-Tier Web Application
Fully automated AWS web application deployment using CloudFormation with VPC, Application Load Balancer, Auto Scaling Group, and EC2 instances across multiple AZs.
View Project DetailsServerless Notes Application
A reliable, low-cost serverless notes management service built on API Gateway + Lambda + DynamoDB with detailed deployment troubleshooting journey.
View Project DetailsS3 Static Website Pipeline
Hands-on AWS learning experience building a static website hosted on S3 and deployed automatically using CodePipeline and CodeBuild.
View Project DetailsCross-Region S3 Replication
AWS S3 CrossâRegion Replication learning journey demonstrating persistence, IAM role configuration, and cross-account/cross-region setup.
View Project DetailsCurrently Building
AWS Solutions Architect Associate
Developing deeper knowledge of AWS architecture, networking, security, resilience and cost optimisation.
Cloud Networking
Building and testing VPC architectures involving public/private subnets, routing, NAT and security controls.
AI Agent Automation
Developing local AI-agent workflows capable of research, automation, scheduled tasks and tool use.
Serverless AI Applications
Continuing to explore Amazon Bedrock, RAG architectures, document intelligence and practical AI integrations.
Technical Skills
AWS & Cloud
AI & Data
Development
Automation & DevOps
IT
Learning Notes & Guides
AWS Lambda CORS Configuration Guide
Learn how to configure CORS settings in AWS Lambda functions to enable cross-origin requests from web applications.
View GuideFind & Remove AWS Credentials on Windows
Securely locate and remove AWS access keys, secret keys, and temporary credentials from Windows systems to prevent accidental exposure.
View GuideAWS Static Website Deployment with CI/CD
Complete guide to deploying static websites on AWS with GitHub Actions CI/CD, including S3, CloudFront, and GitHub Pages integration.
View GuideAWS Security Best Practices
Essential security configurations, IAM policies, VPC designs, and compliance considerations for AWS deployments.
View GuideServerless vs Server-Based Contact Forms
Comparison of serverless and traditional server-based contact form implementations, covering pros, cons, and use cases.
View GuideFull CRUD API Tester
Web-based API testing tool with full CRUD operations (Create, Read, Update, Delete) for REST and GraphQL endpoints.
View ToolMy career combines more than a decade of customer-facing experience with hands-on IT support and a growing portfolio of AWS, AI and automation projects.
My IT support background gave me practical experience troubleshooting technology, supporting users and working through technical problems. I've since been developing those skills further through hands-on AWS projects, cloud architecture labs and AI automation.
I'm now looking to take that combination of technical problem-solving, customer experience and practical cloud development into my next IT or cloud-focused role.