Tony – Python, React, Typescript

A venture-backed AI platform developing frontier LLMs to solve complex software engineering challenges as a step toward AGI.

• Selected as a top open-source contributor to review and label human- and AI-generated code for model fine-tuning;
• Performed prompt engineering to guide models in solving complex software engineering tasks previously handled by human engineers;
• Reviewed code quality, correctness, and architecture across AI- and human-produced solutions;
• Conducted pairwise comparisons to evaluate model outputs and guide training toward higher coding proficiency;
• Contributed high-quality training data used to improve large-scale LLM performance on real-world engineering problems

Achievement: Awarded weekly bonus for top task quality and volume among a cohort of 400+ engineers.

A mobile chatbot application for Auckland residents that answers waste and recycling questions using Azure NLP and Microsoft Bot Framework, helping reduce contamination and improve recycling outcomes.

• Conceived the product idea, pitched it to the council, and led delivery from concept to production;
• Implemented intent classification and entity extraction using Azure NLP to provide accurate, context-aware responses;
• Integrated Microsoft Bot Framework to power conversational flows and backend messaging infrastructure;
• Built and maintained the chatbot platform for over 7 years of continuous service.

A unified web portal combining data exploration, API documentation, developer tooling, and self-service onboarding, built with Remix to support over 400,000 monthly users while driving product adoption and conversion.

• Built modern, high-performance UIs with Remix to showcase Electricity Maps data through interactive explorers and maps;
• Built the portal from the ground up with PostHog analytics to measure user behavior, optimize conversion funnels, and gain deep insights into the end-to-end user journey;
• Drove product-led growth by designing self-service flows and improving the authenticated user experience across 400k+ monthly users.

A high-scale electricity data API delivering real-time, historical, and forecast signals to enterprise customers (including Google, Microsoft, Meta, Samsung, and Cisco), enabling energy trading, smart devices, live dashboards, and carbon accounting worldwide.

• Led the Advocacy team, owning commercial API strategy and product decisions for a high-traffic, customer-facing platform;
• Drove development of a stable, high-performance API serving 160k tokens and ~500M monthly requests with 100% uptime;
• Built custom endpoints for FAANG and enterprise clients to meet specific product and service requirements;
• Balanced enterprise customer needs (including Fortune-scale clients) with large-volume free users via the Home Assistant ecosystem;
• Guided roadmap and prioritization for real-time, historical, and forecasting endpoints to maximize customer impact and platform reliability.

An open-source, global electricity emissions map that uses transparent data pipelines and simple color-based visualization to educate millions of users each month and promote awareness through trustworthy, accessible insights.

• Migrated the legacy codebase from JavaScript and Create React App to TypeScript and Vite, significantly improving developer velocity and reducing time to ship new features;
• Implemented new time granularities across frontend and backend systems;
• Added historical navigation to improve data exploration and user experience;
• Deployed and maintained 1:1 mobile applications built from the same shared codebase.