Resume

Akbar Anvarkhujaev

London · Los Angeles · a.anvarkhujaev@lse.ac.uk · akbruin06@g.ucla.edu

Education

London School of Economics and Political Science

Sep 2026 – Jun 2027

Study Abroad Year · London, UK

University of California, Los Angeles

Sep 2024 – Jun 2028

B.S. Mathematics/Economics, Specialization in Computing · Minor in Data Science · Los Angeles, CA

  • GPA 3.8/4.0 · Dean's Honors List (3 quarters)

Oxford International College

2022 – 2024

A-levels: Mathematics, Further Mathematics, Economics, Russian · Oxford, UK

  • A*A*A*A · Extended Project Qualification: A

Experience

Ipak Yuli Bank

Jun 2026 – Sep 2026

Risk Analyst Intern · Tashkent, Uzbekistan

  • Contributed to the bank's ICAAP (Internal Capital Adequacy Assessment Process) framework and integrated risk-management practice.
  • Improved ICAAP reporting across key risk areas and worked with stakeholders to design new risk metrics.
  • Developed risk-appetite metrics and helped redesign the risk-appetite framework to support management decision-making.

UCLA Department of Mathematics

Sep 2025 – Jun 2026

Undergraduate Research Assistant · Los Angeles, CA

  • Co-authored a NeurIPS 2026 main-track paper on Shapley value estimation (see Research).
  • Developed theoretical and computational approaches that reduce sampling requirements while preserving estimation accuracy.
  • Applied randomized algorithms, leverage-score sampling and sparse recovery.

European Bank for Reconstruction and Development (EBRD)

Jun 2025 – Aug 2025

Credit Analyst Intern · Tashkent, Uzbekistan

  • Analysed sustainability-linked loans against green-finance benchmarks.
  • Built a DCF model to evaluate investment viability.
  • Modelled cost and energy savings for investments in cold-storage facilities and sustainable cutlery production.

Research

Provably Accurate Shapley Value Estimation in High Dimensions via Sparse Leverage Sampling

R. Parada, A. Anvarkhujaev, C. Martin, W. Chang, L. Tran-Thanh, N. Phuong Tran

NeurIPS 2026 · Main Track (Poster)

Shapley values measure how much each input contributes to a model's prediction, but computing them exactly takes exponentially many evaluations. We introduce Sparse-Leverage SHAP, a sampling scheme for settings where only a few inputs matter, and prove it reaches relative-error accuracy with O~(s2)\widetilde{O}(s^2) evaluations, depending on the number of influential inputs ss rather than the total number nn.

OpenReview →

Honors & awards

Leadership & activities

International Finance Student Association

Nov 2024 – Jul 2025

Treasurer · UCLA

  • Led consulting engagements with BNP Paribas, MABA Corporate Services and Picasso Markets.
  • Co-organised the 2025 Harvard Global Case Competition (200+ teams).

Bruin Dine Sustainability Committee

Sep 2025 – Jun 2026

Vice President · UCLA

  • Built a model comparing the carbon footprint of animal-based and plant-based meals.

CalTeach Program

Jan 2026 – Apr 2026

Volunteer Math Instructor · Los Angeles, CA

Skills & interests

Programming
Python (NumPy, pandas, SciPy, Matplotlib), R, C++, SQL, TypeScript / React
Quantitative
Optimisation, statistical modelling, randomized algorithms, backtesting, DCF and credit analysis
Tools
Git, LaTeX, Jupyter, Excel
Languages
English (fluent), Russian and Uzbek (bilingual)
Interests
Chess, competitive GeoGuessr (top 10 in Uzbekistan), football analytics and film.