I build production software and AI systems that are designed to be reliable beyond demos.
My work spans full-stack engineering, AI orchestration, system architecture, long-term memory, semantic retrieval, model routing, cloud infrastructure, and production AI systems.
Note
Most of my recent engineering work has been on commercial products and client systems, so many of my latest repositories are private.
This GitHub primarily contains personal projects and experiments.
For a broader overview of my experience, feel free to visit my LinkedIn:
https://www.linkedin.com/in/giwahenry
If you're specifically interested in the AI systems and architectures I've been designing, I've documented many of them here:
I enjoy solving engineering problems where software architecture and AI meet.
Some of the problems I spend most of my time thinking about include:
- Designing AI systems that combine deterministic logic with LLM reasoning.
- Building production software that scales beyond prototypes.
- Model routing and intelligent orchestration.
- Context engineering and long-term memory systems.
- Retrieval architectures and knowledge systems.
- AI evaluation and benchmarking.
- Distributed systems and cloud infrastructure.
- Building products that remain reliable under real-world usage instead of demo environments.
ChessIQ is an AI chess coaching platform designed around reliable AI systems rather than simply integrating LLMs.
Its architecture combines deterministic chess analysis with conversational AI to deliver personalised coaching while balancing capability, latency, reliability, and cost.
Core architecture includes:
- Deterministic chess analysis (Stockfish)
- LLM-powered coaching
- Model-agnostic routing
- Long-term memory
- Semantic retrieval
- Behavioural pattern recognition
- Context assembly
- Player profiling
- Cost-aware model selection
- AI orchestration pipelines
The engineering philosophy behind ChessIQ is simple:
The quality of an AI product depends far more on the architecture around the models than the models themselves.
Founder of ArcnetLabs, where I'm building practical AI products that solve real-world problems through thoughtful systems design.
Current areas of interest include:
- AI Agents
- Context Engineering
- Long-Term Memory
- AI Orchestration
- Evaluation Frameworks
- Retrieval Architectures
- Human-AI Interaction
- Production AI Infrastructure
Over the last few years I've worked across full-stack engineering and AI systems engineering, taking products from prototype to production while designing the infrastructure and AI architecture behind them.
Some highlights include:
- Leading the stabilisation and production launch of a healthcare platform.
- Designing AI system architectures that combine deterministic reasoning with LLM-powered reasoning.
- Building scalable full-stack applications and cloud infrastructure.
- Designing model-routing systems that balance capability, latency, reliability, and cost.
- Benchmarking frontier AI models using difficult reasoning and coding tasks to expose failure modes and improve evaluation quality.
- AI Agents
- LLM Orchestration
- Model Routing
- Context Engineering
- Long-Term Memory
- Semantic Retrieval
- Retrieval-Augmented Generation (RAG)
- Evaluation Frameworks
- Prompt Engineering
- Node.js
- TypeScript
- Python
- Express
- Flask
- Django
- REST APIs
- WebSockets
- Authentication
- PostgreSQL
- MongoDB
- MySQL
- React
- Next.js
- TypeScript
- JavaScript
- HTML5
- CSS3
- Tailwind CSS
- Material UI
- Chakra UI
- Redux
- React Query
- AWS
- Docker
- Redis
- Firebase
- GitHub Actions
- CI/CD
- Linux
- Git
https://ai-systems-architecture-portfolio.netlify.app/
https://www.linkedin.com/in/giwahenry
📧 Email
henrywilder000@gmail.com
💼 LinkedIn
https://www.linkedin.com/in/giwahenry
🐙 GitHub
https://github.com/CIPHER-000
"Great AI products aren't built by choosing better models. They're built by designing better systems."