I am an AI engineer focused on turning LLMs, agents, retrieval, and automation into systems that can be evaluated, governed, deployed, and trusted.
My current work sits around enterprise AI for operational environments: model routing, tool governance, document intelligence, industrial workflows, eval gates, and AI systems that have to behave under real constraints.
I have also built across healthcare, accessibility, developer tooling, finance, web apps, data visualization, OCR, and AI assistant workflows. The common thread is simple: I like building systems that make messy information usable.
AI systems LLM apps, agents, evaluation, routing, tool use Enterprise focus governance, approval flows, audit trails, reliability Applied domains oil and gas, operations, healthcare, accessibility, finance Delivery style FastAPI, Python, TypeScript, Docker, GitHub Actions, cloud patterns
| Repository | Focus |
|---|---|
| Enterprise AgentOps Control Plane | LLMOps, model routing, traces, cost controls, eval endpoints |
| MCP Gateway for Regulated Tools | governed tool access, RBAC, approval tickets, audit logs |
| Industrial Copilot for Maintenance Ops | industrial triage, risk scoring, human review, KPIs |
| Operational Document Classifier | train/evaluate/serve loop, confidence thresholds, review routing |
| AI Coding Agent Sandbox | coding-agent workflow, diff proposal, approval gate, sandboxed tests |
| LLM Reliability Eval Harness | eval datasets, quality gates, traces, CI regression checks |
| Area | Repositories |
|---|---|
| Agent and developer tooling | graphify, everything-claude-code, openclaw, context-hub, google-cli |
| Retrieval and knowledge systems | RAG-NHS, AI-Agents-Aura-Farm, free-llm-api-resources, public-apis |
| Healthcare and accessibility | uk-accessibilty-advisor, NHS-Capacity-Dashboard |
| Finance and decision tools | StockAgent, Inventory-Optimise, Expense-Tracker, newssummarise |
| Web and product builds | potfolio-website, Shahis-flavour-house, picklewebsite, hackkathon |
| Data science foundations | DataVizStockDataStock, Stock-Analysis-Introduction2DataScience, Hackathon |
AI Engineering LLMs, agents, RAG, evals, model routing, prompt systems Frameworks FastAPI, React, Next.js, Node.js Data and Search Postgres, OpenSearch, JSONL eval sets, structured APIs Reliability tests, traces, quality gates, CI/CD, review workflows Cloud Direction Vertex AI, Azure AI, Docker, GitHub Actions
Calling an LLM is the easy part.
The real engineering work is everything around it:
- measuring whether outputs improved or regressed
- deciding when humans should approve an action
- tracing why a model, tool, or workflow step was chosen
- turning experiments into APIs, tests, dashboards, and deployment paths
- making systems useful in domains where mistakes have consequences
- AI Engineer roles
- Applied AI and forward-deployed AI roles
- LLMOps and AI platform engineering
- enterprise agent systems
- industrial AI and operational workflow automation