Contact: abismuth@g.harvard.edu • linkedin.com/in/alex-bsmth

Download PDF
Education

Harvard University — Cambridge, MA, USA · 2026 – 2027
MS Data Science

  • Coursework: Data Science, Machine Learning, MLOps, Modern AI Systems, Distributed Training & Inference, Capstone Project
  • MIT Cross-registration: How to Make (Almost) Anything, Sensorimotor Learning, Embodied Intelligence

University of Oxford — Oxford, UK · 2025 – 2026
MSc in Advanced Computer Science, Artificial Intelligence • 76% (Distinction)

  • Coursework: Geometric Deep Learning, Uncertainty in Deep Learning, Computer Vision, Game Theory, Database Systems

École Polytechnique — Paris, France · 2022 – 2025
BSc in Mathematics & Computer Science • GPA 3.8/4.0 (Cum Laude)

  • Coursework: Linear Algebra, Analysis, Statistics, Stochastics, Machine Learning, Algorithms, Comp. Architecture, Concurrency
  • International Honors Program, Stanford University: Tech Entrepreneurship and Artificial Intelligence • GPA 3.7
  • Visiting student, University of Oxford: Mathematics and Computer Science • GPA 4.0
Experience

Graduate Researcher — British Open-Ended Learning & Discovery Lab, UK · Apr 2026 – Present

  • Building large market LOB foundation models for high-frequency trading co-supervised by Prof Ani Calinescu and Prof Jakob Foerster
  • Sped up training by >2x with Pallas kernels and improved LOB-Bench market realism by 20% via a novel GAN post-training method
  • Probed state-space models, showing market-microstructure features are linearly decodable from activations and causally steerable

Undergraduate Researcher — University of Oxford, UK · Jan 2025 – Mar 2025

  • Bachelor thesis on gunshot detection in tropical forests for wildlife protection with TinyML supervised by Prof Alex Rogers
  • Developed a SoTA ML Arduino pipeline with 256kB RAM and 1MB memory via model compression and knowledge distillation

Software Development Engineer Intern — Amazon, Spain · Jul 2024 – Sep 2024

  • Designed and implemented the migration of Amazon Charts’ backend architecture over three months, reducing costs by $10k/year
  • Shipped features for Amazon Books, deploying ‘Best Sellers’ reports to South Africa/Poland and creating new targeting filters
Publications

Adversarial Post-Training of Foundation Models for High-Frequency Trading · ICAIF 2026
A. Bismuth*, K. Li*, A. Marti-Shahandeh, V. Mohl, J. Foerster, A. Calinescu · *Co-first authors

Understanding the Inner Workings of a State-Space Foundation Model for Limit Order Books · ICAIF 2026
V. Mohl*, A. Bismuth*, A. Marti-Shahandeh, G. Nigmatulin, K. Li, S. Zohren, J. Foerster, M. Cucuringu, A. Calinescu

Why don’t Graph Transformers attend to the same token? · GDDL Workshop, NeurIPS 2026
A. Bismuth

Scaling Laws for Order Flow Generation · Under Review, ICLR 2027
K. Li, A. Marti-Shahandeh, V. Mohl, S. Y. Frey, G. Nigmatulin, A. Bismuth, L. Sánchez-Betancourt, S. Zohren, A. Calinescu, M. Cucuringu, B. Sarkar, J. Foerster

Awards

Anne Dias Griffin and Kenneth C. Griffin Fellow — Harvard University

Leadership

Treasurer — Duo Conseil X-HEC · 2023 – 2025

  • Oversaw a student consulting society as board member, generating $60k of ARR

President — X’Srael · 2022 – 2023

  • Led a fully funded TechTrip enabling 42 students to interact with Israel’s startup ecosystem
Skills & Interests
  • Programming — Python, C++, C, CUDA, Kotlin, JavaScript, Scala, Haskell, SQL || PyTorch, JAX, TensorFlow, AWS, HPCs
  • Languages — French (Native), English (Fluent — 116/120 TOEFL), Spanish (Fluent)
  • Interests — Music: Singing, Piano, Guitar || Sports: Endurance Running, Motorsport, Rowing, Padel, Rugby