Curriculum Vitae
↓ Download PDFResearch Interests
Anomaly Detection, Self-supervised/Unsupervised machine learning (ML), Fast/"edge" ML, Foundation Models, Beyond Standard Model Physics, Future Colliders.
Education
Ph.D. candidate in Physics — Stanford University
Sep 2023 – Present · Stanford, CA · Advisors: Dr. Julia L. Gonski, Dr. Dong Su
B.Sc. in Mathematics and Physics (Honors in Major) — University of Wisconsin-Madison
Sep 2019 – May 2023 · Madison, WI
Senior Honor Thesis: "Prospects for the Measurement of the Standard Model Higgs Boson Pair Production at the Muon Colliders." Advisor: Dr. Sridhara R. Dasu.
Experience
Research Assistant — SLAC National Accelerator Laboratory
June 2023 – Present · Menlo Park, CA
ATLAS experiment at the LHC
- Interpretable anomaly detection for new physics searches: Developed an interpretable two-stages framework combining supervised/self-supervised contrastive learning with a downstream anomaly detection task. Introduced a factorized-likelihood interpretation enabling hypothesis testing and signal extraction with calibrated significance. Demonstrated significant gains in sensitivity over baselines across diverse signatures.
- ORCA for dedicated new physics searches: Applied the ORCA framework for long-lived particle searches, using contrastive learning to detect extremely rare decay anomalies that standard physics algorithms miss, then using a data-driven background estimate method to extract the signal with a calibrated uncertainty.
- Real-time anomaly detection at the L0Global calorimeter trigger: Adapted ORCA to calorimeter-tower images via a lightweight MLP-Mixer backbone for FPGA deployment under stringent resource budgets. Improves performance over existing trigger-level baselines while inheriting ORCA's interpretability, enabling post-hoc diagnostics of trigger decisions in real-time AD systems.
- Ultra-fast ML on FPGAs at the ATLAS experiment: Led the design and deployment of the experiment's first operational, unsupervised anomaly detection trigger on custom FPGAs. Processing 40 MHz data streams within strict <25 ns latency, the system leverages pure-data training to capture novel physics signals previously discarded by standard algorithms.
- ML Deployment Infrastructure: Stationed at CERN, Genève (Summer 2025), developed a C++ framework for hls4ml (High Level Synthesis for ML) to deploy models into the ATLAS real-time data filter simulation. It generates wrappers around the ML model and produces pre-compiled shared libraries to automate data preprocessing and interface between the detector response and the neural network.
Hardware-aware AI/ML and Microelectronics
- Embedded Sensor Intelligence: Collaborated with electrical engineers to embed ML directly onto pixel sensors (eFPGAs) to manage bandwidth constraints. Optimized a Boosted Decision Tree (BDT) for extreme power budgets (milliwatt level) with single-clock cycle latency (O(10) ns).
- Data Compression & Detector Monitoring: Evolved the system to use a Variational Autoencoder at the sensor edge. This architecture reduces data transmission by 80–90% while simultaneously effectively identifying sensor defects (e.g., dead or "loud" pixels) via anomaly scoring.
Undergraduate Research Assistant — University of Wisconsin-Madison
May 2020 – May 2023 · Department of Physics · Madison, WI
- Worked under Dr. Sridhara Dasu on feasibility studies of measuring di-Higgs production and Dark Matter Searches at the future Muon Collider. Both papers are included in the Snowmass 2021 proceedings, part of the U.S. particle physics planning process held every decade.
- Proposed a novel approach of simulating the background noise induced from beam effect at the future Muon Collider within a fast Monte Carlo simulation framework.
Selected Publications
Variational autoencoders for at-source data reduction and anomaly detection in high energy particle detectors
A. Yue, H. Jia, J. Gonski · Machine Learning: Science and Technology 6(3):035017 · 2025
Analysis of hardware synthesis strategies for machine learning in collider trigger and data acquisition
H. Jia, A. Dave, J. Gonski, R. Herbst · arXiv:2411.11678 · 2024
Embedded FPGA developments in 130 nm and 28 nm CMOS for machine learning in particle detector readout
J. Gonski, A. Gupta, H. Jia, H. Kim, L. Rota, L. Ruckman, A. Dragone, R. Herbst · Journal of Instrumentation 19(08):P08023 · 2024
Muon Collider Forum report
K. Black, et al. · Journal of Instrumentation 19(02):T02015 · 2024
Prospects for the Measurement of the Standard Model Higgs Pair Production at the Muon Colliders
K. Black, et al. · Proceedings of the US Community Study on the Future of Particle Physics (Snowmass 2021) · 2023
Report of the Topical Group on Physics Beyond the Standard Model at Energy Frontier for Snowmass 2021
T. Bose, et al. · Proceedings of the US Community Study on the Future of Particle Physics (Snowmass 2021) · 2022
Summarizing Experimental Sensitivities of Collider Experiments to Dark Matter Models and Comparison to Other Experiments
A. Boveia, et al. · Proceedings of the US Community Study on the Future of Particle Physics (Snowmass 2021) · 2022
Prospects for Heavy WIMP Dark Matter Searches at Muon Colliders
K. Black, et al. · Proceedings of the US Community Study on the Future of Particle Physics (Snowmass 2021) · 2022
Talks and Presentations
GELATO: A Generic Event-Level Anomaly detection Trigger for ATLAS
Fast Machine Learning for Science 2025 · Zürich, Switzerland · Sep 2025
Firmware:ML integration into Athena
ATLAS & CMS ML Operations Workshop · Genève, Switzerland · Jun 2025
Integrating HLS4ML into Athena
ATLAS Trigger and Data Acquisition (TDAQ) Week · Genève, Switzerland · Mar 2025
eFPGA-based ML Implementation on Future Collider Detector Readout
US Higgs Factory Planning 2024 · Menlo Park, CA · Dec 2024
eFPGA-based ML Implementation on Future Collider Detector Readout
US LHC Users Association Annual Meeting · Menlo Park, CA · Dec 2024
Edge Machine Learning for Smart Detectors in Future Colliders
Bridging the Farm: AI for Science at SLAC and Stanford · Stanford, CA · Oct 2024
Embedded FPGA Developments in 130/28 nm CMOS for ML in Detector Readout
US ATLAS Summer Workshop 2024 · Seattle, WA · Jul 2024
Prospects for the Measurement of the Standard Model Higgs Pair Production at the Muon Colliders
Fermilab Users Meeting 2023 (Poster) · Batavia, IL · Jun 2023
Physics Prospects at the Muon Collider
UW-Madison Physics Board of Visitors Meeting · Madison, WI · May 2023