Visiting Student - HEP - Lalich, Ava - 7.9.26

This position is closely connected to a new AI-enabled common near-far inference concept proposed through the Genesis Mission framework, in which the DUNE near detector (ND) and far detector (FD) are treated as a unified scientific inference system rather than as largely separate detector problems linked primarily through simulation-based transfer.

DUNE is a flagship DOE HEP experiment whose scientific reach depends on efficiently translating detector information into validated oscillation analyses. The scale of the experiment—millions of channels in the ND and four large FD modules—makes it a compelling target for HPC-enabled AI acceleration. Current workflows are limited by imperfect visible-to-true energy inference, detector-response differences between the ND and FD, heavy reliance on simulation-driven near-to-far transfer, calibration complexity, and the separation between reconstruction and oscillation-fit stages.

Our proposed common near-far framework addresses this bottleneck by learning a shared, multimodal, fit-aware latent representation optimized not only for reconstruction, but also for downstream oscillation analysis under uncertainty.
We will prototype and benchmark this workflow using simulated ND and FD datasets, with truth information where needed and prototype/calibration information where useful for robustness studies. Accordingly, the project will focus on developing and testing transformer-based common near-far inference models that couple ND and FD event information in a shared latent space. This approach has the potential to accelerate and strengthen DUNE oscillation analyses, including future CP violation measurements.
 

Education and Experience Requirements

• Must be currently enrolled full-time at an accredited institution.

• Must be 18 years or older at the time the appointment begins.

Ava Lalich is a strong candidate who brings the background needed to undertake this project.

The student should understand the basics of neutrino event reconstruction using AI/ML-based deep-learning reconstruction packages such as SPINE. This background is important because the project requires access to reconstructed neutrino events together with their corresponding final-state particles.

Information will be extracted from SPINE HDF5 files, including neutrino event flavor and energy, event-level charge hits and their charges and coordinates, and track- and shower-level hits with their associated charges and coordinates. These inputs will be provided to a transformer network designed to classify neutrino type and regress neutrino energy.

The network will be trained on a high-statistics DUNE Near Detector event sample and then applied to a lower-statistics DUNE Far Detector neutrino event sample through a common near-far inference layer currently under development.
On the implementation side, the student will work within a Jupyter/Python environment to develop code for preparing data inputs for transformer training and inference models, running the models on HPC platforms such as Polaris at ANL, and analyzing and presenting results.

Ava has already received training in these topics as a recent Metcalf Fellow at ANL (Summer 2025).

Job Family

Visiting Student Undergraduate

Job Profile

Visiting Student - Undergraduate

Worker Type

Contingent Worker

Time Type

Part time

Scheduled Weekly Hours

20

EEO Information

As an equal employment opportunity employer, and in accordance with our core values of impact, safety, respect, integrity and teamwork, Argonne National Laboratory is committed to a safe and welcoming workplace that fosters collaborative scientific discovery and innovation. Argonne encourages everyone to apply for employment. Argonne is committed to nondiscrimination and considers all qualified applicants for employment without regard to any characteristic protected by law.

Argonne employees, and certain guest researchers and contractors, are subject to particular restrictions related to participation in Foreign Government Sponsored or Affiliated Activities, as defined and detailed in United States Department of Energy Order 486.1A. You will be asked to disclose any such participation in the application phase for review by Argonne's Legal Department.