I mentor and consult for people and groups in industry and academia.

Are large language models necessary? Do we have large data or small data? How do you validate an AI model with human-in-the-loop feedback? If you are deciding whether machine learning may even help in a given context, I can help. I’ve written code to analyze hundreds of gigabytes of data generated by millions of people to find patterns, make predictions, and help stakeholders make decisions at scale.

Previous projects include:

  • For The Browser (140k+ users), developed a large language model-enabled newsfeed to make the editorial process more efficient.

  • With data from the Lose It! app (40M users), built a food recommendation system that scales to millions of users and tens of millions of datapoints.

  • For a metabolomics lab, helped build performant and interpretable machine learning solutions to identify molecules in biological samples from mass spectrometry instruments. Applications include drug discovery, blood, stool, biomarker, and food analysis.

  • For arXiv.org data, developed state-of-the-art recommender system described in Chapter 4 of my thesis to analyze the reading behavior of hundreds of thousands of users across 500k+ papers to identify trends and insights.

  • Contributed initial UI/UX design research and prototypes for CANImmunize, which was used for the roll-out of Canada’s COVID-19 vaccine.

  • For clinical notes in hospital electronic health records, helped develop a large language model to predict readmission that has 600+ citations and has been used across several academic medical centers.

Feel free to reach out if you think we might work together: [email protected]. I’m especially interested in helping use my knowledge of AI & machine learning to improve products using state-of-the-art methods such as the scalable large language models and recommender systems I have built.