AI
Real-time AI for particle physics
Machine-learning systems that help identify scientifically interesting collision events before data is filtered away.
Hello, I’m Katya
Katya Govorkova
Particle physicist working at the intersection of AI, real-time anomaly detection, and hardware-accelerated scientific discovery.
She develops machine-learning systems for high-energy physics, with a focus on real-time event selection at the Large Hadron Collider, unsupervised anomaly detection, and neural-network deployment on ultra-low-latency hardware. Her work explores how AI can help scientific instruments identify rare signals inside enormous streams of data.
Research focus
Real-time AI for scientific discovery.
Katya’s work sits between experimental particle physics, machine learning, hardware-aware model design, and open scientific benchmarks.
⌁
Anomaly detection for new physics
Unsupervised models that look for statistically unusual events without relying on predefined labels.
▣
Machine learning on FPGAs
Neural-network deployment on specialized hardware for ultra-low-latency inference in trigger systems.
↔
Scientific datasets and open benchmarks
Public resources that let the research community test LHC-style anomaly-detection algorithms.
Featured publications
Selected Nature Portfolio work.
Autoencoders on field-programmable gate arrays for real-time, unsupervised new physics detection at 40 MHz at the Large Hadron Collider
Why it matters: Demonstrates that deep-learning anomaly detection can run inside LHC-style trigger systems under extreme latency constraints.
LHC physics dataset for unsupervised New Physics detection at 40 MHz
Why it matters: Provides a public benchmark dataset for developing and comparing unsupervised new-physics detection algorithms.
Research explained simply
Why this work matters.
?
Why does the LHC need AI?
The LHC produces far more collisions than can be stored. AI can help decide which events are worth keeping.
◎
What is anomaly detection?
Instead of searching only for known signatures, anomaly detection asks whether an event looks unusual compared with ordinary data.
⚡
Why hardware matters
In real-time physics systems, a good model is not enough. It also has to run fast enough on physical electronics.
Talks, lectures, and media
AI, physics, and hardware acceleration.
Her research connects particle physics, artificial intelligence, and specialized hardware, making it relevant to scientific conferences, AI seminars, hardware-acceleration workshops, and public discussions about AI for science.
◉
Invited talks
Conference talks and seminars on real-time machine learning for particle physics.
⌬
Lectures and schools
Educational material on anomaly detection, trigger systems, and scientific AI.
✦
Media inquiries
Public-facing explanations of how AI can help scientific instruments discover rare signals.
Projects / collaborations
From collider triggers to open benchmarks.
CMS
Real-time ML for collider triggers
Machine-learning systems for identifying interesting particle-collision events under strict latency and bandwidth constraints.
FPGA
Hardware-aware scientific AI
Efficient inference and model deployment for experimental physics systems.
DATA
Open benchmarks
Datasets and evaluation tools that help researchers compare anomaly-detection methods.
ML
Scientific representation learning
Embeddings, flows, and machine-learning methods for downstream discovery tasks.
Academic CV
A clear professional record.
A full CV page can include current position, previous positions, education, awards, selected grants or fellowships, teaching and supervision, service, reviewing, committees, and a complete publication list.
Positions
Current and previous institutional roles.
Education
PhD, degrees, fellowships, and academic training.
Service
Teaching, committees, reviewing, and community contributions.
AI systems that help scientific instruments discover rare signals in impossible amounts of data.