Projects

Built a multi-stage AI pipeline that transforms raw survey data into evidence-based SWOT analyses. The application ingests CSV exports and automatically extracts themes, identifies patterns, and generates strategic insights with supporting evidence.

Features built-in PII protection, fault-tolerant processing with checkpoint/resume capability, and generates actionable "How Might We" questions. Reduced a 2-day manual consulting process to under 2 hours with consistent, data-driven outputs.

Ongoing coaching and mentorship for small businesses and educators on practical AI adoption. Topics include prompt engineering and context engineering strategies, AI workflows, using AI for research, and evaluating AI tooling options.

Sessions also cover AI risks and opportunities specific to each business, AI adoption strategies, developing AI policies, and building long-term AI strategy. Clients include small businesses across industries and the education company behind the Use Case Generator project.

Built a headless FastAPI backend powered by Claude Agent SDK with multi-vendor AI pipelines. Exposes unified REST/SSE endpoints for chat, deep research, image generation, voice transcription, text-to-speech, and knowledge base features.

The API-first architecture enables hybrid Claude + Gemini workflows with persistent session management. Includes an extensible library of skills, productivity commands, and specialized agents—all accessible through a single authenticated API.

Built a multi-agent workflow system on Google's Agent Development Kit (ADK) that transforms user input into polished use case documentation. Specialized agents handle clarification, research, brainstorming, outlining, writing, and editing—working together to produce consistent, high-quality outputs.

Features autonomous web research, video shot logging for instructional content, and auto-publishing to GitHub. Includes a FastAPI backend with real-time streaming and a web UI with invite-code authentication.

Built a library of self-contained agent skills for TaskFox Factory using progressive disclosure design. Each skill bundles instructions, automation scripts, and reference docs that agents load on-demand—extending AI capabilities for specialized tasks without overwhelming context windows.

Developed multi-session orchestration enabling complex projects to span multiple agent sessions with state management. Used this approach to build complete applications—backend, frontend, and deployment—entirely through coordinated agent sessions in sandboxed cloud environments.

Used AI to develop comprehensive assessment materials for UK apprenticeship programs focused on AI preparedness. Created multiple-choice quizzes, project-based assessments, and supporting resources including datasets and solution guides—all aligned with industry-relevant AI competencies.

Collaborated with a leading UK apprenticeship education provider to build assessments measuring learners' progress in AI skills from prompt engineering to data management. Applied AI-assisted research and prompt engineering throughout development to create accurate, job-relevant evaluation materials.

Building a professional caregiver training course for ALS (amyotrophic lateral sclerosis) on the LearnWorlds learning management system. The 5-unit curriculum covers foundation knowledge, activities of daily living, quality of life, communication and support—with ~180 lessons drawn from expert-reviewed source content.

Video narration uses ElevenLabs text-to-speech with configuration-driven voice generation. Provided extensive consultation on LMS selection, course curriculum structure, instructional design, and revenue/monetization strategy.

Built voice selection and audio generation tools using ElevenLabs TTS for healthcare educational videos. The system evaluates multiple voice candidates across different style variations—from warm and conversational to measured and clinical—with tunable parameters for fine-grained control.

Dual quality tiers enable cost-effective iteration: a fast preview mode for rapid drafts and a studio mode for production-ready audio. Configuration-driven generation ensures reproducibility across sessions. The client compared dozens of voice samples before selecting the final narrator for their video production.

Created "Prompting 101" for a boutique branding and marketing agency. The 11-chapter reference guide covers prompt fundamentals through advanced techniques including Chain-of-Thought, Few-Shot Learning, and XML-structured prompting.

Addresses 5 agency service areas: Brand Development, Marketing Strategies, Web & UX Design, Creative Production, and Research. Includes real client case studies and prompt templates for campaign briefs, creative briefs, and 4 audit types (brand, website, content, marketing).

Formulated comprehensive curriculum for four online courses teaching all GitHub Copilot skills for a leading apprenticeship education provider. Courses cover the complete Copilot capability set—from code completion and chat to test generation and code explanation—with hands-on projects building real developer proficiency.

Built all course content and projects to rigorous educational standards using advanced prompt engineering and context engineering techniques. Applied systematic AI-assisted workflows to ensure technical accuracy while maintaining pedagogical effectiveness across all four courses.

Built a RAG-based conversational AI with a full-stack web interface for domain-specific Q&A. The system combines a modern frontend with a FastAPI backend, using vector storage for semantic search across the client's knowledge base.

Ingested hundreds of website pages and video transcripts from the client's existing content. Users receive AI responses with source attribution and compliance disclaimers. Features resilient error handling and native authentication. Deployed to production cloud hosting.

Built a lightweight frontend for the TaskFox Factory API using Alpine.js and vanilla JavaScript. The single-page application features real-time chat streaming, voice input/output, session management, and intuitive discovery interfaces for skills, agents, and commands.

Zero build tooling or npm client dependencies—just modular ES6 code. Features include model selection, research mode toggle, image generation, and an events timeline. Serves as our in-house R&D playground for testing new TaskFox Factory capabilities.

Built a Claude-backed Amazon Bedrock AgentCore runtime that bridges AWS's managed AI infrastructure with the Claude Agent SDK. The serverless deployment handles conversational AI agents with integrated memory, identity management, code execution, and browser automation capabilities.

The runtime integrates AWS Bedrock's memory services for conversation persistence across containers, plus code execution and web automation tools. Includes comprehensive test coverage and least-privilege IAM policies for secure production deployment.