Fruit of Preterition

Charles Renshaw-Whitman Email | Phone: +1 (609) 217-2136 | LinkedIn | GitHub | Personal Website and Blog

Trained in mechanical engineering (McGill, honors) and applied physics / sustainable energy technology (TU Delft, dual MSc., cum laude). Self-taught the additional mathematics - real analysis, differential geometry, information geometry, convex optimization - to transition into foundational ML theory research. Currently a MATS scholar under Richard Ngo, working on the structural properties of natural data that enable learned intelligence.

RESEARCH INTERESTS

I study the structure of natural data and the structured computation it induces in learners. Specific current directions include: operationalizing world-model comparison in POMDPs via spectral methods (Hankel matrices); developing principled mathematical tools to deconfuse concepts like ‘circuits’ and ‘predictive salience’; and characterizing universal phenomena in deep learning using techniques from asymptotic analysis and statistical physics.

RESEARCH EXPERIENCE

MATS Scholar Foundational Theory of Learnability and Data Structure Remote, Under Richard Ngo January 2026 – Present

  • Researching foundational question: what structural properties of data enable learned intelligence.
  • Developing theoretical framework using tools from information geometry to disentangle data structure from learning dynamics.
  • Applying spectral methods (Hankel matrices) to operationalize world-model comparison in POMDPs.
  • Interim research notes available at website.

SERI MATS Scholar Infra-Bayesianism for People Who Don’t Know Measure-Theory: Remote, Under John Wentworth June 2022 – September 2022

  • Studied recent theoretical work on the problem of aligning superintelligent AI systems to human values.
  • Self-taught convex analysis and measure-theory to produce an accessible distillation of ‘Infra-Bayesianism’, a modified probability theory designed to allow pessimization over outcomes of non-modelable processes.

EleutherAI Summer of AI Research (SOAR) Improving Automated Interpretability Remote, Under Gonçalo Paulo August 2025

  • Developed pipeline for iterative refinement of natural-language explanations of SAE latent activations.
  • Contributed to open-source Delphi automated-interpretability package.

MSc. Applied Physics Thesis Electrostatic Modelling of Quantum Dot Arrays Technische Universiteit Delft, Netherlands January 2022 – August 2022

  • Developed polynomial-time optimization-based solution to quantum-dot tuning problem.
  • Analyzed numerical methods to solve the inverse problem of electrostatic system identification.
  • Explicated mathematical framework for control of electron populations on quantum dots.
  • Results published in SciPost Physics Codebases (v.i.).

MSc. Sustainable Energy Technology Thesis Reinforcement Learning Methodology for Electricity Market Design Technische Universiteit Delft, Netherlands January 2023 – August 2023

  • Developed a solution to the non-stationarity problem for reinforcement-learning-based evaluation of electricity-market designs.
  • Proposed a novel reinforcement-learning method for the optimization of electricity-market design constrained by actors’ strategic behavior.
  • Results published as first author in Electric Power Systems Research (v.i.).

Honors Thesis Research Dynamics of Multiheaded Waves in the Rotating Detonation Engine McGill University, Canada January 2020 – April 2021

  • Developed CFD software to simulate detonative processes with application to the Rotating Detonation Engine.
  • Invented reduced-order sub-grid model for reactant mixing.
  • Proposed and evaluated an algebraic model governing the formation of wave instabilities.
  • Performed verification of existing CFD codebase.
  • Presented research at the AIAA conference with a first-author publication.

EDUCATION

Technische Universiteit Delft, Delft, Netherlands September 2021 – August 2023 GPA: 8.51 of 10.0

  • Degrees: MSc. in Applied Physics (cum laude); MSc. in Sustainable Energy Technology earned simultaneously within the allotted two years
  • Relevant coursework: Statistical Learning Theory, Deep Reinforcement Learning, Quantum Information (RL-based quantum circuit design)
  • Leadership Roles:
    • President, TU Delft Debating Society (2022-2023)
    • Secretary, TU Delft Debating Society (2021-2022)

McGill University, Montréal, Canada September 2017 – April 2021 GPA: 3.74 of 4.00

  • Degree: B.Eng. in Honors Mechanical Engineering with an Honors Track minor in Physics
  • Honors Thesis: Dynamics of Multiheaded Waves in the Rotating Detonation Engine
    • Podium presentation and first author of paper presented at AIAA Conference.

INDEPENDENT STUDY (a small selection)

  • Deep Learning Theory: Worked through Principles of Deep Learning Theory (Roberts et al. 2023) and Deep Learning (Goodfellow et al. 2016) with exercises and notes.
  • Mathematical Foundations: Self-taught real analysis, topology, differential geometry, information geometry, convex optimization, and stochastic differential equations beyond formal coursework.
  • ML Alignment Engineering: Completed ARENA mechanistic interpretability curriculum; OpenAI “Spinning Up in Deep RL”.
  • Representation Learning: Audited MILA graduate course (Representation Learning).

PROFESSIONAL EXPERIENCE

Machine Learning Researcher, Makena AI (f.k.a. Neuron3D) July 2025 – December 2025

  • Built commercial product using Gaussian splatting to generate 3D models for property management.
  • Designed synthetic data pipeline using Nvidia’s Difix LoRA to enhance model quality.
  • Performed end-to-end design and analysis of machine learning workflow to ensure performance and quality.
  • Studied current 3D ML literature to inform key architectural decisions.

Operations Director, Del Buono’s Bakery September 2023 – June 2025

  • Oversaw financial operations, auditing cash flow for four major retail locations.
  • Coordinated payroll for over 140 employees.
  • Advised the company’s leadership during debt restructuring, improving financial sustainability.
  • Oversaw doubling of production in a three week period in response to sudden closure of a competitor.

Propulsion System Lead, McGill Rocket Team September 2019 – June 2020

  • Led the optimization of propulsion systems for a student-researched and designed (SRAD) hybrid rocket engine.
  • Developed a system-level propulsion simulator to analyze the performance of hot-fire tests.
  • Designed test infrastructure for hybrid motor propulsion systems, facilitating accurate measurement of flow properties.
  • Managed cross-disciplinary engineering teams in iterative development of test-site infrastructure.

Engineering Intern, Argospire Medical May 2017 – August 2017

  • Co-ordinated with medical professionals to identify requirements for a novel respiratory medical device.
  • Collaborated to design, prototype, and test design-iterations of the device.
  • Performed numerical simulations of fluid flow to determine performance and measurement characteristics.
  • Advised on integration of device with machine learning algorithms to evaluate user’s breathing and administer medicine accordingly.

RESEARCH PUBLICATIONS

  1. Renshaw-Whitman, C., Zobernig, V., Cremer, J. L., & de Vries, L. (2024). Non-stationarity in multiagent reinforcement learning in electricity market simulation. Electric Power Systems Research, 235, Article 110712. https://doi.org/10.1016/j.epsr.2024.110712
  2. Gualtieri, V., Renshaw-Whitman, C., Hernandes, V., & Greplova, E. (2025). QDsim: A user-friendly toolbox for simulating large-scale quantum dot devices. SciPost Physics Codebases, 46. https://doi.org/10.21468/SciPostPhysCodeb.46
  3. Renshaw-Whitman, C., Mi, X., & Higgins, A. J. (2020). Computational simulation of multi-headed detonation dynamics in rotating detonation engines. AIAA Conference Proceedings. https://doi.org/10.2514/6.2020-3877

SKILLS

  • Mathematics & Physics: Real Analysis, Topology, Differential Geometry, Information Geometry, Asymptotic Analysis, Convex Optimization, Stochastic Differential Equations, Group Theory; Quantum Field Theory, Analytical Mechanics, General Relativity, Phase Transitions, Fibre Bundles, Topological Solitons
  • Programming: Python (PyTorch), MATLAB, C/C++, Linux/Bash

LANGUAGES

  • English: Native
  • Mandarin: Proficient
  • Latin: Proficient
  • Ancient Greek: Proficient (Reading and writing only)
  • Portuguese: Beginner

REFERENCES

Available upon request