Curriculum Vitae

Download as PDF One-page resume (PDF)

Education

PhD in Statistics and Applied Probability, UC Santa Barbara · Sep 2021 – Jun 2026
Advisor: Tomoyuki Ichiba. MA in Statistics, 2026.
Teaching assistant for 12 courses over four years: PSTAT 5A, 5LS, 8, 10, 120A/B, 123, 160A/B, 170, 174/274 and 176/276. Mentored 10+ undergraduate researchers.

BS in Mathematics, UC San Diego · Sep 2019 – Jun 2021
Dean’s Undergraduate Excellence Award (top 0.5%, Physical Sciences).

Research

Noise-State Calculus for Dynamic Games with Strategic Information
PhD dissertation, 2026 · sbabichenko.com/dissertation
Dynamic games in which players learn from each other’s actions and can shape what others believe: continuous-time linear-quadratic-Gaussian games with any finite number of players. The information wedge, the shadow price of changing an opponent’s beliefs, explains where these games depart from their full-information versions. Applications to Kyle–Back markets with several informed traders, networks of local markets, delayed public signals, and deviations that only some players can detect.

Forecasting and Manipulating the Forecasts of Others · arXiv: 2603.12140
Solo-authored, submitted March 2026. The core of the dissertation. In a two-player benchmark, splits the cost of dispersed information into an estimation part and a strategic part, and finds cases where more precise private information raises total cost.

Software. noisestate, a Python package that solves these games from a model description: pip install noisestate. A C++ version runs in the browser explorer at sbabichenko.com/noisestate.

Talks. "Forecasting and Manipulating the Forecasts of Others," Southern California Quantitative Finance Forum (SCQF), UC Santa Barbara, April 2026. "Mean Field Games and Interacting Particle Systems" (following Daniel Lacker), UCSD Stochastic Systems Seminar, January 2021.

Experience

Wells Fargo, Charlotte, NC · Quantitative Researcher, rotational program · Jul 2026 – present

Wells Fargo, Charlotte, NC · Quantitative Researcher Intern, Mortgage Model Development · Jun – Aug 2025

• Built tree-based segmentation of mortgage pools, with diagnostic tools.
• Modeled loan-level prepayment.

Selected Projects

Decision Mesh · sbabichenko.com/decisionmesh · github.com/sbabichenko/triangular-decision-mesh
Adaptive, continuous regression, inspired by decision trees, that refines only where a false-discovery gate finds support, and decomposes the data into a surface, unit-level effects, and noise. The rectangular version cuts one axis at a time and scales to higher dimensions.

Multi-agent traffic simulation (Waymo Open Motion Dataset)
Reimplemented BehaviorGPT, a decoder-only transformer with agent–agent attention, for closed-loop 10 Hz trajectory generation. Extending it so the transformer predicts only at branch points where behavior becomes multimodal and a conditional diffusion model fills in between.

Skills

Programming: C++23, Python, CUDA, SQL, Linux

ML and data: PyTorch, transformers, diffusion models, LightGBM, XGBoost, Polars, NumPy

Methods: stochastic calculus and control, Kalman filtering, Monte Carlo, PDE methods, hierarchical Bayesian models, time series, derivatives pricing

Last updated: September 2026