I'm a software engineer, data scientist, and founder based in Toronto, with 9+ years of experience turning complex ideas into useful software. My background in cognitive science shapes how I approach problems: understanding the people, the underlying data, and the system as a whole.
At the University of Toronto, I lead Dynamic Datasets development for CanDIAPL, building astronomy data infrastructure and tools that help researchers work with large, varied datasets. Previously, I spent six years at Creative Destruction Lab, delivering 50+ software and R&D projects that collectively saved over 10,000 hours of manual work.
My AI experience spans statistical modelling, NLP, and LLM applications. Early work at CDL included transformer autoencoders, Hugging Face, and embeddings; today I build systems that combine language models with structured data, retrieval and validation. My SOILIE-3D research explores computational visual imagination through the generation of 3D scenes.
I also founded Halo Media Solutions, where I take products from concept through architecture, development, and deployment. My projects include AI-powered tools such as Domain Seeker, the Worlds Become game, and applications for music, visual art, and media. That experience shapes how I think about user experience, maintainability, and the practical tradeoffs of running a product.
I work primarily with Python, TypeScript, React/Next.js, SQL, and AWS, though I am comfortable with the entire software stack on any developer platform. I enjoy collaborating across disciplines, turning open-ended ideas into practical plans, and building systems that people can understand, use, and maintain.
AI and language models: Natural Language Processing, Machine Learning, Large Language Models, Hugging Face, Transformers, Embeddings, Model Evaluation
Data and verification: Structured Data Extraction, Data Engineering, SQL, PostgreSQL, Data Validation, Data Provenance, Database design
Applied AI workflows: Retrieval-Augmented Generation, Human-in-the-Loop Workflows, OpenAI, Anthropic, Google Gemini, Amazon Bedrock
Infrastructure and delivery: AWS, REST APIs, Docker, Kubernetes, CI/CD, GitHub Actions, OAuth, OpenID Connect, Technical Documentation
Full-stack breadth: Python, TypeScript, JavaScript, React, Next.js, Flask, Pandas, NumPy, scikit-learn, Cloudflare Workers, PostGIS, TimescaleDB
Technical lead for Dynamic Datasets within CanDIAPL, a national, multi-university astronomy inf...
Technical lead for Dynamic Datasets within CanDIAPL, a national, multi-university astronomy infrastructure initiative.
• Architect data services and ingestion workflows for heterogeneous survey data, including PostgreSQL/PostGIS/TimescaleDB designs, metadata validation and CADC/CANFAR integration.
• Deployed CANFAR services on an Arbutus Kubernetes cluster with shared CephFS storage and CADC authentication/group authorization; successfully ran thousands of headless jobs.
• Develop COMET, SILKEN and Multi-Messenger Portal tooling for software onboarding, deployment, data discovery, research workflows, and user access.
• Assess implementation readiness, identify capability gaps and integration dependencies, and translate findings into architecture documentation, priorities, and operational guides.
• Collaborate with scientists and partner engineers across Canadian research institutions to turn evolving requirements into maintainable services.
Founded and run an independent software company, owning product strategy, architecture, implementation, deployment, and ongoing iterat...
Founded and run an independent software company, owning product strategy, architecture, implementation, deployment, and ongoing iteration.
• Build AI-enabled products and developer tools, including Domain Seeker, and the pre-release Worlds Become LLM-powered MMORPG and a text-to-3D generation platform & API.
• Develop LLM integrations with structured output validation, independent result checks, caching, quotas, usage accounting, and cost controls.
• Implement retrieval-augmented NPC memory and bounded model contexts, with server-side validation of proposed actions.
• Design full-stack applications and cloud services using TypeScript, Python, Next.js, AWS and Cloudflare, including authentication, asynchronous processing, and persistent data models.
• Develop additional media and creative tools spanning image transfer, colour analysis, music composition, and local rendering.
• Own requirements, user experience, technical documentation, and delivery decisions across a portfolio of products at different development stages.
Long-running research collaboration on computational visual imagination and the SOILIE-3D scene-gen...
Long-running research collaboration on computational visual imagination and the SOILIE-3D scene-generation system.
• Develop algorithms and research software for representing, processing, and visualizing labelled 3D scene data and generating novel indoor environments.
• Investigate geometric and data-quality failure modes, compare implementation approaches, and document their implications.
• Built the public SOILIE-3D research website and demonstrator, and extended comparative evaluation and technical reporting.
• Collaborate with researchers on experimental design, implementation challenges, analysis and manuscript revisions.
This is an ongoing research collaboration, separate from my full-time employment.