Connor Fraser-Grant

• Design and deliver AI tooling and full-stack software from requirements and architecture through testing, deployment, and production support.
• Created OrgSDK, a secure multi-tenant runtime for recurring agent workflows with isolated code execution, permission-based tools, durable state, schedules, approvals, and traces.
• Built a model-flexible harness around OpenRouter and open-source models, using bounded tasks, explicit tool schemas, and targeted context to improve smaller-model reliability.

• Conceived, pitched, designed, and implemented Functionly's AI agent, owning the product vision and technical delivery from idea to release.
• Built an agent that read and edited organisational scenarios spanning roles, positions, functions, responsibilities, and hierarchical relationships.
• Engineered the agent loop and harness for self-correcting tool calls, object references, clarification questions, sub-agents, and research grounded in Functionly and organisational design.
• Improved reliability across tool-call accuracy, context growth, cache invalidation, ID handling, and hierarchical-data representation; supplied live interface context so the agent understood what each user was viewing.
• Released the agent to all users; multiple customers upgraded for the complete feature set, with positive feedback that it made large organisational changes easier.

• Built a live-beta platform that turns bounded recurring work into hosted agent workflows with more than 100 tools, schedules, approvals, artifacts, and inspectable run traces.
• Isolated each tenant in its own container environment and each action in a fresh V8 isolate with no direct network access and short-lived, resource-scoped credentials.
• Designed the execution engine, Postgres-backed state, permission-based plugins, workflow creation, resource management, and multi-agent orchestration.

Certificate IV in Web Development. Ongoing self-directed learning through courses, research, and hands-on project work.