cv
This is a description of the page. You can modify it in '_pages/cv.md'. You can also change or remove the top pdf download button.
Basics
| Name | ANDREY BRYUTKIN |
| Label | PhD Candidate in Applied Mathematics, MIT |
| bryutkin@mit.edu | |
| Summary | Applied mathematician and physicist specializing in stochastic processes, time series analysis, Bayesian inference, and uncertainty quantification. ICML-published researcher with expertise in developing UQ methods for large-scale models and dynamical systems. Production-level Python and HPC experience. Seeking quantitative research roles in systematic strategies where rigorous mathematical modeling meets real-world financial data. |
Work
-
2025.06 - 2025.08 Research Intern
MIT–IBM Watson AI Lab
Cambridge, MA
- Designed and evaluated uncertainty-aware metrics for Granite Guardian models, including accuracy rejection and out-of-distribution detection; built scalable evaluation pipelines in Python/JAX/PyTorch
- Implemented gradient-based and representation-based UQ scoring methods; automated experiment tracking, data loading, and benchmarking across diverse datasets
- Collaborated with research engineers to enhance code reproducibility and performance through vectorization, batched I/O, and profiling optimization
-
2022.09 - Present Academic Assistant
ETH Zurich
Remote / Zurich
- Co-authored lecture materials for ``Waves & Electrodynamics'' course; developed comprehensive problem sets and autograding scripts; streamlined LaTeX production toolchain
-
2022.02 - 2022.05 Software Engineer—Optimization & Data Systems
pick8ship
Zurich
- Architected warehouse optimization components and data interfaces; integrated algorithms with production databases and robotics systems
- Enhanced data sampling and preprocessing pipelines, improving stability of downstream optimization processes
-
2020.09 - 2022.07 Teaching Assistant
ETH Zurich
Zurich
- Led exercise sessions (20–40 students), authored solutions, and graded assignments for C++ (2020/2021), Physics II (2020), Mathematical Methods II (2022)
-
2020.09 - 2020.12 Research Assistant
ETH Zurich Particle Physics Group (CERN)
Zurich
- Designed and deployed lab monitoring and data management systems using InfluxDB; implemented analytics dashboards for experimental workflows
Volunteer
-
- Present -
- Present Zurich, Switzerland
-
- Present Zurich, Switzerland
Education
-
2023.09 - Present Cambridge, MA
PhD
Massachusetts Institute of Technology (MIT)
Applied Mathematics
- Probability
- Stochastic Processes
- Statistical Inference
- Convex & Numerical Optimization
- Time Series
- Machine Learning
- PDEs
-
2022.10 - 2023.06 Cambridge, UK
MASt (Master of Advanced Study — ``Part III'')
University of Cambridge
Machine Learning, Probability, Partial Differential Equations
-
2019.09 - 2022.09 Zurich, Switzerland
BSc in Physics, with Distinction
ETH Zurich
Physics
- Exchange Semester: University of Toronto (Fall 2021)
Awards
- 2023–2024
- 2022–2023
Part III International Scholarship
University of Cambridge
- 2022–2023
Christ's College Postgraduate Bursary
Christ's College, Cambridge
- 2019–2023
German Academic Scholarship Foundation
German Academic Scholarship Foundation
- 2020–Present
Hans-Messer-Stiftung
Hans-Messer-Stiftung
- 2021
Teaching Assistant Award
ETH Zurich
Publications
-
2025 Neural Triangular Transport Maps for Sampling in Lattice QCD
NeurIPS AI4Science Workshop (to appear)
Bryutkin, A., Marzouk, Y. (2025).
-
2025 Canonical Bayesian Linear System Identification
arXiv
Bryutkin, A., Levine, M. E., Urteaga, I., Marzouk, Y. (2025).
-
2024 HAMLET: Graph Transformer Neural Operator for Partial Differential Equations
ICML 2024
Bryutkin, A., Huang, J., Deng, Z., Yang, G., Schönlieb, C.-B., Aviles-Rivero, A. (2024).
-
Uncertainty-Aware LLM Probing
(to appear)
Bryutkin, A., Li, A., DeOliveira, J., Rundensteiner, E., Gerych, W., Thost, V.
Skills
| Programming | |
| Python (NumPy, Pandas, JAX, PyTorch, SciPy) | |
| C++17 | |
| Julia | |
| MATLAB | |
| LaTeX | |
| SQL |
| ML/Statistics | |
| Bayesian inference | |
| Monte Carlo methods | |
| time series analysis | |
| signal processing | |
| optimization | |
| graphical models |
| Systems | |
| Linux | |
| Git | |
| HPC (vectorization, parallel I/O) | |
| experimentation toolchains |
Languages
| English | |
| fluent |
| German | |
| fluent |
| Russian | |
| native |
Projects
- 2024 - 2025
Canonical Bayesian Linear System Identification
Research at MIT.
- Developed gauge-invariant Bayesian identification framework for LTI systems; derived canonicalization under GL($d$) orbits and efficient MCMC/VI estimators
- Implementation: JAX/NumPyro/BlackJAX; emphasis on numerical stability, Fisher information structure, and priors near unit roots
- 2024 - Present
UQ for Molecular Dynamics Emulation
Collaboration with NVIDIA.
- Developing uncertainty-aware surrogates for force fields on graphs; adapting transformer operators to physical constraints and OOD shifts
- 2024 - Present
UQ for Quantum Field Theory
Collaboration with Columbia University.
- Implementing transport-map methods and scalable samplers for lattice models; exploiting conditional sparsity and site ordering for linear-time triangular maps
- 2022 - 2023
HAMLET: Graph Transformer Neural Operator for PDEs
Research at University of Cambridge.
- Combined diffusion GNNs with Fourier neural operators for inverse problems on graphs; published open-source implementation and benchmarks