Nico Christianson

I am a fifth-year Ph.D. candidate in Computing and Mathematical Sciences at Caltech, supported by an NSF Graduate Research Fellowship and a PIMCO Data Science Fellowship. My research is broadly concerned with decision-making under uncertainty, with a specific focus on developing new algorithms to enable deploying modern AI/ML tools reliably and safely to real-world problems in energy and sustainability.

In my work, I have developed optimal algorithms to leverage black-box AI/ML advice in online optimization while ensuring provable, worst-case performance guarantees. I have also worked on designing such “learning-augmented” algorithms in a variety of more general, constrained problems. Recently, I’ve been particularly interested in studying how to design better algorithms for risk-aware decision-making, especially by using uncertainty quantification.

Beyond my theoretical work, I am deeply invested in translating theoretical insights to practical, real-world impact, and I have worked on a variety of applications including electricity markets and carbon-aware datacenter operation. During my Ph.D., I collaborated with Beyond Limits to apply our learning-augmented algorithms to improve cogeneration management in high-renewables power grids, and I led the implementation of their cogeneration system model in the open-source RL benchmark suite Sustaingym. I have also collaborated with Amazon Prime Video on developing and deploying new, theoretically-grounded algorithms for adaptive bitrate streaming.

At Caltech, I am fortunate to be advised by Adam Wierman and Steven Low. During the summer of 2023, I interned at Microsoft Research Redmond with Weiwei Yang and Baosen Zhang. Previously, I received an A.B. in applied mathematics from Harvard, where I worked with Boris Kozinsky on ionic transport, and with Dani S. Bassett (UPenn) on network science.

Check out some of our recent collaboration/deployment work:

Preview image
SODA: An Adaptive Bitrate Controller for Consistent High-Quality Video Streaming
Preview image
SustainGym: An RL Benchmark Suite for Sustainable Energy Systems

selected publications (full list here)

  1. AISTATS
    Optimal Robustness-Consistency Tradeoffs for Learning-Augmented
    Metrical Task Systems
    N. Christianson, J. Shen, and A. Wierman
    International Conference on Artificial Intelligence and Statistics, 2023
  2. COLT
    Risk-Sensitive Online Algorithms
    N. Christianson, B. Sun, and 2 more authors
    Conference on Learning Theory, 2024
  3. arXiv
    End-to-End Conformal Calibration for Optimization Under Uncertainty
    C. Yeh*, N. Christianson*, and 3 more authors
    Under review, 2024
  4. arXiv
    Fast and Reliable N−k Contingency Screening with Input-Convex Neural Networks
    N. Christianson, W. Cui, and 3 more authors
    Under review, 2024
  5. SIGMETRICS
    The Online Pause and Resume Problem:
    Optimal Algorithms and An Application to Carbon-Aware Load Shifting
    A. Lechowicz, N. Christianson, and 5 more authors
    ACM SIGMETRICS/IFIP Performance, 2024
  6. ICML
    Online Algorithms with Uncertainty-Quantified Predictions
    B. Sun, J. Huang, N. Christianson, and 3 more authors
    International Conference on Machine Learning, 2024
  7. NeurIPS
    SustainGym: Reinforcement Learning Environments for Sustainable Energy Systems
    C. Yeh, V. Li, R. Datta, J. Arroyo, N. Christianson, and 7 more authors
    Neural Information Processing Systems Datasets and Benchmarks Track, 2023

news

Sep 30, 2024 I’m pleased to have been named a 2024 PIMCO Graduate Fellow in Data Science!
Sep 18, 2024 Excited to discuss my work on reliable AI/ML for energy systems and sustainability in three upcoming invited talks:
  • 10/10/24 - Cornell ORIE Young Researchers Workshop
  • 10/18/24 - UMass Amherst Sustainability Seminar
  • 10/21/24 - INFORMS Annual meeting, 10:45am (Session MB70 “Optimization and Learning for Sustainable Grids”, Regency - 701)
Jul 21, 2024 We have two papers at ICML this week on online algorithms with UQ predictions and convex function chasing with long-term constraints.
Jul 3, 2024 I am giving the Amii AI Seminar at the University of Alberta. Watch the recording here.
Jun 30, 2024 I am at COLT in Edmonton this week to present on our recent work on Risk-Sensitive Online Algorithms.
Jun 14, 2024 Congratulations to our outstanding undergraduate researchers James, Helen, and Jerry on their graduation from Caltech! Wishing them all the best as they begin their doctoral studies!
Jun 10, 2024 I am at ACM SIGMETRICS in Venice this week! We have two papers in the main conference on new (learning-augmented) algorithms for carbon-aware datacenter operation, and I will also be giving talks in our Learning-augmented Algorithms workshop and the MAMA workshop.
Jan 10, 2024 Excited to be co-organizing the second annual workshop on Learning-augmented Algorithms: Theory and Applications at ACM SIGMETRICS/IFIP PERFORMANCE 2024!
Oct 24, 2023 I am giving a talk at the UMass Amherst Theory Seminar.