Hi, I'm
Mila Bileska

New physics may be hiding in the rarest signals; finding it is a challenge of inference, computation, and precision that I pursue relentlessly.

About Me

ORCID | LinkedIn | GitHub

Advancing discovery at the intersection of physics and technology.

Ph.D. student in Physics at Princeton University, awarded the merit-based Centennial Fellowship in the Natural Sciences for doctoral study. Specializing in high-energy experimental physics, with a focus on firmware development, ML-enabled trigger systems, and hardware-accelerated data processing for beyond-the-Standard-Model searches and Higgs studies at the Compact Muon Solenoid (CMS) experiment. Graduate of Princeton with a Bachelor of Arts in Physics and Minor in Computer Science, conferred with Honors in Physics and awarded the Allen G. Shenstone Prize in Physics. Affiliated with CMS at CERN and the U.S. Muon Collider Collaboration.

SELECTED PUBLICATIONS

Reconfigurable On-Chip AI for Particle Physics Detectors with Logic Neural Networks
• Venue: NeurIPS 2026 Workshop on On-Device Intelligence

Design and FPGA Implementation of WOMBAT: A Deep Neural Network Level-1 Trigger System May 2025
for Jet Substructure Identification and BoostedH → b¯b Tagging at the CMS Experiment
• Publisher: Journal of Instrumentation

A review of non-Lorentz invariant variable speed of light theories August 2024
• Publisher: Classical and Quantum Gravity (IOP Science)

Spectroscopy and modeling of 171Yb Rydberg states for high-fidelity two-qubit gates June 2024
• Publisher: Physical Review X

At the heart of my work is the quest to uncover beyond the Standard Model processes: from the development of real-time intelligent trigger systems to searching collision data for anomalous signals, I have actively explored the interplay between scientific research and technology.

My CV

EXPERIMENTAL PHYSICS DOCTORAL STUDENT

Programming Languages

C, C++, C#, Java, Python, JavaScript, HTML, CSS, ARM Assembly, Vivado HLS, LaTeX

Simulation Tools

Geant4, MadGraph5_aMC@NLO, Powheg, SIMION, Fusion 360, AutoCAD

Firmware

Field Programmable Gate Arrays

Libraries

[Firmware: Xilinx Vitis, hls4ml, PetaLinux][Physics Research: CMSSW, alpaka, FastJet, Coffea, GSL,
LLAMA, MadGraph, mplhep, PyROOT, ROOT, SIMION, Uproot, awkward][Machine Learning: Keras,
QKeras, LightGBM, MXNet, PyTorch, Scikit, TensorFlow, XGBoost] [Data Processing and Visualization:
Graphviz, Matplotlib, OpenCV, Pandas, Pydot, SciPy, Tabulate, h5py]

Languages

English (C2), Macedonian (native), Croatian (B2), Serbian (B2), Bosnian (B2), Bulgarian (B2), Montenegrin
(B2), French (A2), Slovenian (A2)

Experience

AND PROJECTS

Developer and Researcher of Logic Neural Networks

- Princeton University

January 2026 – Present

• Project: Developing an open-source PyTorch library for training Logic Neural Networks (LNNs), differentiable models that learn Boolean logic functions and compile into discrete logic-gate representations for efficient, hardware-friendly inference.

• Role: Core developer

• Contributions: Implemented model features and developed a firmware-testing interface for RTL conversion, hardware validation, and FPGA synthesis of trained LNN models.

• Principal Investigators: Prof. Isobel Ojalvo, Dr. Lino Gerlach

Developer and Researcher of Logic Gate Networks

Project: Logic Neural Networks for SmartPixels Applications

- CMS CERN, Fermilab, Princeton University

September 2025 - Present

• Publication: NeurIPS

• Project: Developing reconfigurable Logic Neural Network architectures for real-time, on-detector inference in future particle-physics pixel detectors under stringent latency, power, and hardware-resource constraints.

• Role: Project lead and main developer

• Contributions: Demonstrated 19× lower FPGA resource utilization than quantized neural networks at comparable accuracy; achieved >0.84 balanced accuracy with convolutional LNNs; and developed reconfigurable LNNs retaining 99% of retrained accuracy at 37% of the resource cost.

• Principal Investigators: Prof. Isobel Ojalvo, Dr. Lino Gerlach

SmartPixels Researcher and Developer

Developer of FPGA Firmware and Emulator for the CMS GCT Sum Card and MHH Module

- CMS, CERN

September 2025 – Present

• Project: Develop and test firmware and emulator software for the GCT Sum Card and MHH Module, components of the Phase-2 CMS trigger system’s FPGA network.

• Role: Project lead and main developer

• Contributions: Implemented FPGA firmware and emulator logic for real-time trigger processing; used custom routing to deploy and test the design in hardware (APx board).

• Principal Investigator: Prof. Alexander Savin

Developer of GCT Sum Card and MHH Module

ML-Based Trigger System Developer for H->bb̅ Detection

- CMS, CERN

January 2024 - May 2026

• Publication: JINST TH 001

• Project: Developed machine-learning algorithms for the CMS Level-1 Trigger, targeting boosted H->bb̅ jet tagging, QCD background rejection, and anomaly detection.

• Role: Project lead and main developer.

• Contributions: Led the development, CMSSW evaluation, and FPGA implementation of WOMBAT, a boosted-jet clustering and triggering algorithm.

• Principal Investigator: Prof. Isobel Ojalvo

WOMBAT Developer and Evaluator

Intern at Princeton’s Quantum Computing and Error Correction Laboratory

- Princeton University

June 2023 - August 2023

• Publication: Physical Review X

• Project: Conducted ¹⁷¹Yb Rydberg spectroscopy to support development of a refined multichannel quantum defect theory model for high-fidelity two-qubit gates.

• Role: Co-researcher.

• Contributions: Performed experimental spectroscopy, data collection, and quantum-defect modeling.

• Principal Investigator: Prof. Jeff Thompson

Published Pre-University Research in the Journal of Physics and Astronomy