About me
I am a (final year) PhD candidate (with Tamara Broderick) in the EECS department at MIT affiliated with LIDS and a junior investigator in IAIFI. Previously I got two MSc degrees in Statistics and Ecology from the University of Wisconsin-Madison and a ScM degree from MIT in Computer Science. I did my undergraduate degree in Biology/Physics at Peking University.
I am primarily interested in probabilistic machine learning and statistics for low signal-to-noise and data-scarce settings, where incorporating structure, designing data collection strategies, and carefully managing downstream risk are essential. My work is broadly motivated by scientific problems in biology and physics that naturally operate in this regime, ranging from perturbation experiments in gene regulatory networks and intracellular fluid dynamics to wildlife population surveys and the calibration of standard candles in the cosmic distance ladder.
I’m currently on job market! I am looking for postdoc, TT assistant professor, research scientist or quantitative researcher roles. Please feel free to email me if you want to chat!
Current projects
- Trajectory and SDE inference in computational biology using multimarginal Schrödinger bridges and least square in distribution space; identifiability theory of stochastic dynamics;
- Theory of learning human preference from binary annotations, corresponding experimental design and robustness checks, with applications in Large Language Model alignment and evaluations;
- Probabilistic machine learning and generative models in physics, especially time domain astronomy, focusing on methods for irregularly sampled sequence data (e.g., timeseries, spectra, etc.) and simulation based inference;
- Population ecology of apex predators and capture-recapture models (e.g., jaguars and wolves).
Some stuff I have worked in the past
- Species and feature sampling problem with heterogeneity, multivariate (completely) random measures and application in genetics and sequencing strategies;
- Identifying stellar flares from photometric data using Hidden Markov Models;
- Spike-and-Slab LASSO on multivariate regressions/chain graphs, its frequentist properties and experimental design for microbiome studies.
More about me
I am a wildlife photographer and angler in my spare time. I hold a technician class amateur radio license, bearing KD9TZJ call sign. I am also an ACG fan.