This job has expired on Indeed
Reasons could include: the employer is not accepting applications, is not actively hiring, or is reviewing applications

Postdocs in Data Intensive Scientific Machine Learning

University of California, San Diego
4.2 out of 5
San Diego, CA
From $60,000 a year - Full-time


San Diego, CA

Pulled from the full job description

  • Dental insurance
  • Flexible schedule
  • Health insurance
  • Parental leave
  • Relocation assistance

Full job description

We seek exceptional postdoctoral candidates to be hosted at UCSD/HDSI, Columbia, or UCI. These positions have an initial term of one year with the possibility of extension. The starting date is flexible, but no later than Fall 2022.


The past two decades have witnessed natural disasters and extreme weather events that affect millions of lives. At the same time, the data volume from high-resolution climate models, satellite, in-situ and ground-based measurements have substantially increased to petabyte scales. These new and readily accessible datasets create the previously missing pipeline for scientific machine learning (ML), which in turn can improve our understanding and ability to predict extreme climate events.

This project will develop deep latent variable models (LVMs) to discover hidden physical structures in high-dimensional, spatiotemporal data of extreme climate events such as droughts or heatwaves. Our project has three research aims:

  • Aim 1: Discover hidden structures from climate extremes: apply novel deep LVMs, such as sequential variational autoencoders (VAEs) and tensor LVMs, to climate simulation data (and particularly to data composed of sparse observations). These LVMs will be designed to extract hidden low dimensional structures that are evolving in space and time. In turn, these hidden structures will be analyzed within the framework of causal transportability, so that predictions can be transferred from simulations to real-world observations in order to more accurately predict those extreme events.
  • Aim 2: Incorporate physical principles, laws, and constraints into deep LVMs and the hidden structures that they discover. Neural networks will be trained by innovative methods that disentangle latent representations, provide weak supervision, and/or enforce equivariance.
  • Aim 3: Quantify the uncertainty of predictions from deep LVMs. Specific goals are to (i) develop scalable techniques for Bayesian inference in deep LVM models, (ii) design pre-conditioners for Monte Carlo sampling procedures that handle long-tailed distributions, (iii) exploit hidden structures to accelerate methods for variational inference (VI), obtain more robust estimates, and improve statistical calibration, and (iv) investigate hybrid methods that combine Monte Carlo sampling and VI for more accurate posterior inference.


  • Ph.D. in computer science, physics, mathematics, engineering, or other related fields.
  • Strong interests in machine learning and earth science, especially deep learning, dynamical systems, spatiotemporal statistics and optimization.
  • Prior experience working with large-scale data sets. Papers in top machine learning conferences and/or scientific journals are a plus.
  • Communication, presentation, writing, and teamwork skills.


  • Curriculum vitae (including publications)
  • Cover letter stating the motivation, interests, and qualifications for the position.
  • Research statement covering past and future research and showing interest and qualifications in the areas relevant to this project (up to 5 pages).
  • Names and contact information of 3 references.

Job Type: Full-time

Pay: From $60,000.00 per year


  • Dental insurance
  • Flexible schedule
  • Health insurance
  • Parental leave
  • Relocation assistance


  • Monday to Friday


  • Doctorate (Preferred)


  • Python: 1 year (Preferred)
  • SQL: 1 year (Preferred)

Work Location: Multiple Locations