Custom GPTs are configurable chatbots tailored through structured instructions, curated knowledge files, and response protocols to support specific domains and tasks. This project develops a generalizable framework for designing pedagogically aligned GPTs as effective AI course companions. As a case study, we present ALISSTAIR, an AI companion for Stats 10: Introduction to Statistical Reasoning. The central problem is how to systematically design instructions, organize knowledge resources, and structure response behavior so GPTs can function reliably in domain-specific settings. We propose a modular framework built on context specification, knowledge design, role/persona definition, instructional guidelines and constraints, and standardized response patterns. Building on prior implementation and refinement, we are piloting ALISSTAIR across two Stats 10 courses enrolling approximately 200 students each, or 400 students total, demonstrating both feasibility and scalability in a large-enrollment undergraduate setting. Ultimately, this work envisions AI course companions as a transformative model for expanding equitable, high-quality academic support across disciplines, institutions, and diverse learner communities.

Accurate, non-intrusive temperature measurement is vital in a range of both scientific and industrial applications. Infrared spectroscopy provides a practical method for obtaining these measurements. However, reliable temperature measurement requires precise calibration of the spectrometer to account for variations in the detector and the system response. In this study, a grating-based monochromator is calibrated using a high-temperature carbon cavity blackbody radiator. Spectral intensity measurements are recorded over a range of measured temperatures, and the Instrument Response Function (IRF) is calculated by comparing the measured spectra to theoretical blackbody radiation calculated from Planck’s radiation law. An optical chopper is designed and implemented to modulate the incident radiation, for improved signal-to-noise ratio. Other spectrometer modifications include installation of a stepper motor and limit switches to allow digital control of the grating angle. The dependence of the IRF on blackbody temperature is examined to evaluate the reliability of the calibration process. The goal of this work is to calibrate an existing spectrometer for use measuring the cathode temperature of the Large Plasma Device (LAPD). Keywords: Spectrometer, Instrument Response Function, Blackbody, Infrared Radiation Calibration

Quantum materials are sensitive to external fields. Recently, our group has developed methods for combining scanning tunneling microscopy/spectroscopy (STM/STS) with in situ uniaxial strain and in-plane electric fields for tuning the properties of quasi-2D van der Waals crystals such as NbSe2. Our approach enables the application of large (~10%), spatially inhomogeneous in-plane strain to bulk crystals, approaching their yield strength limits. Using STM/STS, we observed new electronic behaviors in NbSe₂. Here, we extend these studies to ambient conditions, focusing on how strain modifies optical response and surface structure. To quantify these effects, we combine optical reflectance spectroscopy/microscopy with atomic force microscopy (AFM). Together, these complementary techniques provide a comprehensive picture of the structural and electronic evolution of layered superconductors under strain.

Energy–energy correlators (EECs) are observables that provide a sensitive probe of both perturbative and nonperturbative dynamics in relativistic heavy-ion collisions. Jet–medium interactions enhance particle multiplicity within the jet cone, which must be properly accounted for when extracting the EEC of jet shower hadrons in experiments. To address this issue, we develop an augmentation method that exploits momentum conservation between the near-side and away-side regions, using gamma–jet events with 0–10% centrality in Pb+Pb collisions at center of mass collision energy per nucleon = 5.02 TeV simulated with the CoLBT-hydro model. This approach yields an experimentally reconstructed EEC that shows improved agreement with the EEC of hadrons originating primarily from jet parton splittings. By comparing EECs for jets selected under different constraints in Pb+Pb and p+p collisions, we further investigate medium-induced modifications of jet fragmentation. The methods developed in this study can help test the scenario in which jet fragmentation occurs predominantly outside the quark–gluon plasma.

Muography is an emerging imaging technique based on detecting muons, subatomic particles produced by cosmic rays. In academic settings, muography has proven to be cheaper, safer, and better at penetrating bulk shielding compared to X-ray methods. However, the technique is slow, limiting widespread adoption. We present a comprehensive muography system that uses machine learning to reduce imaging times without sacrificing image quality. An apparatus of eight particle detectors, called multi-wire proportional chambers, is designed and built to collect imaging data. Each detector comprises the gas-tight physical chamber, high-voltage circuitry to produce strong electric fields, and signal processing electronics. Software infrastructure enables data acquisition and remote control. To carry out the experimental procedure, objects are placed at varying positions inside the apparatus, and the resulting muon trajectory distributions are recorded. This process generates paired data (object placement configurations and their corresponding muon trajectory patterns) used to train a supervised learning model to predict object locations from observed trajectories. Faster muography removes the primary barrier to adoption, enabling applications previously impractical at scale, from systematic scanning for concrete defects in bridges to non-invasive imaging of culturally significant monuments and heritage sites.