Data Analyst Machine Learning

Linda Zhenyu Jin

805-***-**** # *********@********.*** § github.com portfolio website

Education

University of Washington, Seattle September 2025

PhD, Physics Seattle, WA

University of California, Santa Barbara September 2020–December 2023 Bachelor of Science, Physics Minor, Comparative Literature Santa Barbara, CA

• Major GPA: 3.93/4.00; GPA: 3.90/4.00; High Honors (Top 8.5%), Dean’s Honors, 2023 Worster Research Fellowship Experience

GenAI Research Data Analyst September 2024–Present Professor Uros Seljak, The Berkeley Center for Cosmology Physics (BCCP) Department of Physics, UC Berkeley

• Improve field-level cosmological inference with generative model (GenAI) by building a conditional CNN in Pytorch for high-resolution hydrodynamical simulations.

• Configure kernels and loss functions with physics constraints in both Eulerian and Fourier space.

• Stack and augment 3D Astrid, SIMBA, IllustrisTNG hydrodynamical simulations for map-to-map training.

• Compare generation performance with conditional Diffusion Model variations, Gaussian Processors, and Normalizing Flows, while leveraging open-source models from Hugging Face and GitHub.

• Deployed data and model pipelines on remote HPC systems (NERSC Perlmutter) to handle large-scale computation.

• Appointed as affiliated scientist at the Lawrence Berkeley National Lab. Astrophysics-ML Research Assistant April 2022–September 2024 Professor Joseph F. Hennawi, ENIGMA Group Department of Physics, UCSB

• Developed and deployed a first production-level machine learning solution to extract thermal information of the intergalactic medium at the post-Reionization epoch. Established automated training and evaluation pipelines with GitHub version control to ensure reproducibility.

• Utilized hyperparameter optimization to achieve a 0.5% emulator error, integrating remote HPC resources through NERSC to scale model training and continuous integration of commits.

• Implemented NumPyro Hamiltonian Monte Carlo for Bayesian inference, achieving fast and accurate parameter estimation. Adopted best practices in software engineering including version management and code reviews via GitHub and Copilot.

• Engineered an automated uncertainty propagation mechanism that passed credibility tests for out-of-distribution data, reducing 17M GPU-hour consumption and ensuring a production-level robust inference pipeline. Publication & Talks

Jin, Z., Wolfson, M., Henna, J. F., & Gonz alez-Hern andez, D. (2024). Neural network emulator to constrain the high-z IGM thermal state from Lyman-α forest flux auto-correlation function. Monthly Notices of the Royal Astronomical Society. https://doi.org/10.1093/mnras/stae2741.

Northwestern University Prof. Claude-Andr e Faucher-Gigu`ere’s Group Meeting Online January 2025 The University of Chicago Prof. Nick Gnedin’s Cosmology Group Meeting Online November 2024 2023 Worster Summer Research Fellowship Awardee Presentation UCSB November 2023 Conference for Undergraduate Women in Physics University of California, Merced January 2023 2022 KITP Undergraduate Physics Research Symposium Kavli Institute for Theoretical Physics September 2022 Leadership & Extracurricular

VP of Finance The Women’s Network, UCSB October 2022–June 2023

• Directed financial planning for a 30+ member chapter with data-driven decision making to drive scalable fundraisers. Peer Advisor College of Letters & Science Academic Advising, UCSB April 2022–June 2023

• Mentored and advised students on degree planning. VP of Events UCSB Chinese Students and Scholars Association August 2020–February 2022

• Led large-scale digital and in-person events, including an online streaming project reaching over 4,000 participants, demonstrating agile project management and the use of social media to drive community engagement. Skills & Related Coursework

• Machine Learning, JAX,

Pytorch, Tensorflow,

Optuna, Hugging Face

• GitHub, Supercomputer

operation, Python,

Jupyter, Fortran

• AWS, Amazon

SageMaker, ZenML,

Comet ML, Docker

• LATEX, Presentation,

Graphic design, Canva,

Figma

• Quantum Mechanics

• Cosmology, Gravitation

and Relativity

• Statistics, Data Analysis,

and Machine Learning

for Physicists

Contact this candidate