profile picture

Tristan B

- Research Program Mentor

PhD at Stanford University

Expertise

Climate change, data science, AI, machine learning, ecology

Bio

I am passionate about using data and AI to address critical issues facing our society and the planet. My research sits at the interface of climate science and statistics (AI/ML). I have investigated a range of environmental systems, with projects exploring climate change impacts on water quality, the California drought, wildfires, and even duck migration patterns. My undergraduate research explored Bayesian statistical approaches to summarizing climate model data. While most or my work is now done online, I have spent time in the field as well--Ask me about measuring ocean acidification at sea for 28 days straight! I also enjoy competing at datathons (think hackathons for data scientists) and racing with Stanford's bike team. I graduated the PhD program in 2021 and have since been working at an AI+climate/weather/sustainability startup.

Project ideas

Project ideas are meant to help inspire student thinking about their own project. Students are in the driver seat of their research and are free to use any or none of the ideas shared by their mentors.

AI for Climate Change

Recent years have seen an explosion of advancements in AI, and the climate and environmental sciences have only begun to adopt these latest methods. Several avenues for projects exist based on student interest in learning certain AI methods (computer vision, natural language processing (NLP), time series methods with transformers) and the climate domain of interest (wildfire, heatwaves, ecosystem impacts, climate justice). Some potential examples include: 1) Using computer vision models to classify wildfire smoke plumes from satellite imagery 2) Using computer vision 'super-resolution' models to enhance the spatial resolution of climate forecasts 3) Time series forecasting of variables like temperature or water levels 4) Natural language processing (NLP) models like BERT applied to classifying climate change misinformation in text data Learning outcomes would include 1) Experience doing an end-to-end data science project on real-world data 2) In-depth knowledge of a particular AI model/algorithm, 2) Python coding experience implementing, training, and fine-tuning the model. A variety of project outcomes are possible, with the most common being a research paper. Prerequisites: Some experience with Python programming. Prior AI coursework experience not expected. Likely outcome is an advanced research paper that can be submitted to high school research journals.

Modeling local climate change impacts

This project explores expected climate change impacts on the student's hometown or any other location of particular interest. The learning objectives include an understanding of the fundamentals of climate science and the tools used to predict future impacts. I am happy to guide the student in framing an engaging topic and to support them in obtaining relevant data, such as historical weather data and predictions out to 2100 from climate models (e.g. from NASA). Specific research directions can be catered to the student's interest, for example placing greater emphasis on statistical analysis or instead diving deeper into policy frameworks (e.g. Paris Climate Accord) and public health dimensions. Students are welcome to think outside the box when considering climate change impacts, which can have a range of secondary effects. For example, droughts have been linked to major conflicts and international migration, sea level rise can exacerbate income inequality, and warmer temperatures can increase the spread of mosquito and tick-borne disease. Potential project outcomes could range from research papers to data analyses to interactive data visualizations. Likely outcome ranges from beginner to advanced research paper, with the advanced route being capable of submission to high school research journals.

How has climate change affected the likelihood of extreme events?

When a natural disaster hits, it is now common to ask whether climate change played a role. From Hurricane Sandy in 2012, to recent heatwaves across Russia, to the massive Great Barrier Reef coral bleaching event, understanding and quantifying the influence of climate change, if any, is increasingly relevant to scientists, local communities, and policymakers. Scientists approach tackling these types of questions through a framework known as detection and attribution, a framework that is currently at the cutting edge of climate science. In short, for a given extreme event, one compares the likelihood of that event occurring in two worlds: the first world being as close to reality as we know, and the second world representing a fictional planet where the industrial revolution essentially never occurred. By comparing the odds of the extreme event in these two worlds, you can begin to quantify the influence of climate change. Fortunately, a degree in climate modeling or computer science isn't necessary to explore these questions. Relevant datasets from climate models can be analyzed in software like Python, R, or Excel. I can help provide this data, or if the student is interested in gaining additional Python experience I can provide example code. The primary learning objectives are 1) Research the various drivers of a particular extreme event, 2) Explore how climate change may influence these drivers, 3) Understand the detection and attribution framework, as well as its controversies, and 4) Gain experience analyzing data and telling a compelling story through data science and data visualizations. Likely outcome is an advanced research paper that can be submitted to high school research journals.

Endangered or invasive species modeling in response to a changing climate change

One of the greatest ecological impacts from climate change will be the shifting of habitats suitable to different species. Of particular concern are shrinking habitats for endangered species, and expanding habitats for invasive species. In this project, a student may select a particular animal/plant species of interest. We will then identify a candidate dataset of locations the species has been observed, such as iNaturalist's database. Then the student can pair that data with information on the typical climate of that region, putting it into what is referred to as a 'species distribution model', a subset of machine learning with the simplest model version being a logistic regression but also including options like random forest. Optionally, the student can then take this trained model and see how expected climates in 2050 and beyond lead to different model predictions of the habitat range. Skills built on in such a project include species distribution modeling, an important area of ecology, as well as in depth research into a specific species. Programming skills for this project require some familiarity with Python or R. Likely outcome is an advanced research paper that can be submitted to high school research journals.

Will the world become too warm for the winter Olympics?

Climate change is increasingly posing significant threats to the Olympics, challenging the tradition of global sports competition. Increasing temperatures and unpredictable weather patterns can disrupt outdoor events. Winter Olympics face particular risks, as warming temperatures jeopardize reliable snow and ice conditions, potentially leading to cancellations or relocations of events. In this project, students will begin by researching the current literature on vulnerabilities of various sports to weather and climate phenomena. Next, the student can compile a list of all previous and/or future planned Olympics locations. We can then see what the projected changes from climate models are in temperature and other relevant variables discovered in your readings to see whether these cities could remain suitable Olympics venues in the years to come. For a more beginner research paper level, the research can focus on readings and data analysis such as through Microsoft Excel. A more advanced research paper level would include working with some of the raw climate model data in Python and more detailed data analysis and data science approaches.

Coding skills

Python, R

Credentials

Work experience

Sust Global (2021 - Current)
Data Scientist

Education

Duke University
BS Bachelor of Science (2011)
Statistics
Stanford University
MS Master of Science (2020)
Statistics
Stanford University
PhD Doctor of Philosophy (2021)
Earth System Science

Reviews

"Working with Tristan was an absolutely wonderful experience. Tristan helped me learn coding in R and data science (from scratch), then apply it to analyse data for my project, and finally complete my research paper, all within 3 months (8-9 sessions). Working with Tristan has been, by far, one of the best academic experience that I have ever had. He made understanding complex topics simple, the approach to writing long research papers natural and easy (as compared to strenuous and difficult), and learning -- FUN. With his insightful feedback, curated assignments, and extremely thorough knowledge eventually outdo my expectations for my Polygence project, and also gain more confidence with Data Science. He was very flexible with scheduling, very understanding when it came to my other commitments (like test prep, college app), and super easy to communicate with through and between my sessions. Tristan also inspired me to spark personal growth and find more clarity with my career goals. He made sure that I achieved all my goals for my Polygence Project, and always gave me extra time post our general discussions to make sure that my Polygence learning experience could great as possible. I feel fortunate to have worked with Tristan."

Arihant from Kolkata, India

Arihant from Kolkata, India profile

Interested in working with expert mentors like Tristan?

Apply now