At The Botss, I lead AI architecture and hands-on development for client projects across legal, healthcare, e-commerce and enterprise. I’ve shipped 9+ production AI systems, including multi-agent assistants, real-time voice agents and RAG applications.
For Arthur, I built a multilingual voice agent serving around 3,000 users across production instances. I introduced prompt caching that reached a 98.5% hit rate and reduced inference costs by ~58%, and designed sub-agents to replace a 6,000+ line monolithic prompt.
I also built the AI agent and backend for Accident Assist AI, which interviews US accident victims, creates structured case files and routes qualified leads to law firms. For RedactIQ, I developed document classification and PII redaction for Australian council planning documents, with inference pinned to the Vertex AI Sydney endpoint.
At Neural Internet, I designed a decentralised LLM chatbot with RAG on the Bittensor network and maintained subnet validators. I also built a real-time validator monitor from node logs that alerted the team on Slack and contributed to a 2% APY increase within weeks.
