Curriculum Vitae

Maximilian Du
maxjdu@stanford.edu / maximiliandu.com / github.com/MaxDu17

EDUCATION

Stanford University Sept 2020—Present

Bachelor of Computer Science (AI Track) + Creative Writing Minor (Prose) + Psychology Minor
GPA: 4.103 / 4.0

Fayetteville-Manlius High School Sept 2016—Jun 2020

High School Diploma
GPA: 103 / 100

PUBLICATIONS

• Maximilian Du et al. “Behavior Retrieval: Few-Shot Imitation Learning by Querying Unlabeled Datasets”. In: Robotics:
Science and Systems XVIV. Robotics: Science and Systems 2023. July 10, 2023. URL: https://arxiv.org/abs/
2304.08742
• Maximilian Du et al. “Play It by Ear: Learning Skills amidst Occlusion through Audio-Visual Imitation Learning”. In:
Robotics: Science and Systems XVIII. Robotics: Science and Systems 2022. June 27, 2022. URL: https://arxiv.
org/abs/2205.14850
• Maximilian Du. “Improving LSTM Neural Networks for Better Short-Term Wind Power Predictions”. In: 2019 IEEE
2nd International Conference on Renewable Energy and Power Engineering (REPE). 2019, pp. 105–109
• Anonymous. “BridgeData V2: A Dataset for Robot Learning at Scale”. In: (June 9, 2023). URL: https : / /
openreview.net/forum?id=f55MlAT1Lu

RESEARCH PROJECTS

Try, Try Again: Behavior Cloning for Novel Test-Time Scenarios Feb 2023–Present
Advised by Sasha Khazatsky, Chelsea Finn & Tobias Gerstenberg
• Created a strategy search algorithm that allows robots to try new strategies upon task failure

Behavior Retrieval: Few-Shot Imitation Learning by Querying Unlabeled Datasets Feb 2022–Jan 2023
Advised by Suraj Nair, Chelsea Finn & Dorsa Sadigh
• Created a novel data selection algorithm that successfully reused a previously-collected dataset of robot interactions to
boost behavior cloning by over 35%
• Ran hundreds of experiments in MuJoCo simulation and created custom real Widowx robot arm control flow to run over
1000 evaluation trials in various real environments.

Play it by Ear: Learning Skills amidst Occlusion through Audio-Visual Imitation Learning Jan 2021—Jan 2022
Advised by Suraj Nair, Chelsea Finn
• Demonstrated that audio data could augment visual and proprioceptive data to improve success rates in certain tasks, like
extracting keys from a bag
• Used MuJoCo, Robosuite, and PyTorch to run reinforcement learning & behavior cloning algorithms in simulation and
on a Franka-Emika Panda robot

Curriculum Vitae | Last Updated 6-18-2023 Page 1

 • Proposed an encoder architecture that incorporates audio spectrogram data
• Developed a data pipeline for Oculus Quest demo collections that other researchers later adopted.

Looking Under the Hood of DetectGPT Mar 2023

CS 224N Final Project
• Verified published results on DetectGPT, an algorithm that discriminates text written by large language models.
• Proposed and tested ways of improving DetectGPT by focusing on certain parts of speech
• Demonstrated that DetectGPT can be partially fooled by an adversarial prompt
• Expanded the results of DetectGPT to include ChatGPT outputs

Can you Macgyver It? Teaching an Agent to Use Tools Mar 2023
CS 234 Final Project
• Implemented a policy gradient algorithm to solve a tool-usage environment
• Explored impacts of different exploration algorithms on data efficiency and final performance

Sixteen Pixels is (Almost) All You Need: Crafting Parameterized Image Uncrumpling Models Jun 2022
CS 231N Final Project
• Modified the Pix2Pix algorithm to take in a crumpled image and output its uncrumpled form
• Proposed a smaller PatchGAN architecture that qualitatively outperforms existing PatchGAN architectures
• Procedurally generated crumpled images using Python and Blender (a 3D rendering engine) to procure a large training
dataset efficiently

MidiStyle: Parameterized Audio Style Transfer for Instrument Swapping Nov 2021
CS 229 Final Project
• Used a convolutional autoencoder to transform piano music into other instruments
• Used a FiLM-style information injection to determine the output instrument

Improving LSTM Neural Networks for Better Short-Term Wind Power Predictions Jun 2018—Nov 2019
Advised by Joshua Comden, Zhenhua Liu
• Demonstrated that a modified LSTM can be used to accurately predict short-term wind power outputs
• Proposed new metrics to measure performance of time series models
• Collected, processed, and combined large (3TB) databases of wind power and weather forecast data

Curriculum Vitae | Last Updated 6-18-2023 Page 2

TECHNICAL AND ACADEMIC SKILLS

Languages and Libraries: Python, PyTorch, Tensorflow, Numpy, Matplotlib, Pandas, C++, C, HTML
Codebase/API Familiarity: ROS, Robosuite, Robomimic, Roboverse, MuJoCo, PyBullet, Oculus Quest, SLURM
Machine Learning: Imitation Learning, Meta-Learning, Reinforcement Learning, Computer Vision, Natural Language
Processing
Math: Probabilistic Graphical Models, Stochastic Processes, Information Theory, Real Analysis, Linear Algebra
Psychology: Learning & Memory, Perception, Cognitive Neuroscience, Behavior Analysis
Tools: Franka-Emika Panda Robot, Widowx Robot, Git, Unix, LATEX, Terminal, Zotero, Adobe Illustrator / Photoshop /
Premiere Pro, Autodesk Inventor, THT/SMD Hand Soldering, Oscilloscope, Spectrum Analyzer, Arduino
Other Skills: Narrative Interviews (30+ hours), Audio Production, Educational Presentations, Creative Non-Fiction, Short
Fiction, Leadership

HONORS AND AWARDS

• 3rd place winner of the Stanford Undergraduate Creative Writing Prize 2023 (out of 500+ submissions)
• CS231N Final Project Winner Spring 2022 (out of 370+ students)
• Finalist of the 2022 Stanford Lunsford Award for Oral Presentation of Research
• CS109 Final Project Winner Spring 2021 (out of 200+ students)
• National Regeneron 2020 Science Talent Search Scholar
• Discovery Education 2020 “Making for Good Challenge” National Second Place
• Finalist of Intel International Science & Engineering Fair, with various special awards in 2018 and 2019

EXPERIENCES

CURIS Participant and IRIS Lab Researcher Jan 2021—Present

Stanford Artificial Intelligence Laboratory (SAIL)
• Worked in the Intelligence through Robotic Interaction at Scale (IRIS) Lab under Prof. chelsea Finn
• Led multiple projects under the supervision of Sasha Khazatsky, Suraj Nair, Prof. Chelsea Finn, Prof. Dorsa Sadigh, and
Prof Tobias Gerstenburg
• Hosted group meetings for undergraduates
• Presented papers in whole-lab reading group

CS 106A/B Section Leader (TA) Jan 2021—Jun 2022

Stanford Computer Science Department
• Led weekly instructive “sections” for the popular CS106A/B Stanford course series. Answered conceptual questions and
guided students through coding problems. Also helped grade assignments and exams.
Stanford Splash Lecturer Nov 2021—Present
Stanford Splash
• Gave lectures to high school students on the connections between animal training and reinforcement learning.

Research / Social Media Advisor Mar 2023—May 2023

Truth4Toki Advocacy Group
• Helped bring national attention to a group of 25 animal trainers (Good Morning America, NBC Seattle, Miami Local 10)
• Helped gain more than 40k signatures on Change.org

Curriculum Vitae | Last Updated 6-18-2023 Page 3

 • Ran organization website and advised social media situations team through a time of controversy

Producer & Writer Jun 2022—Present

Stanford Storytelling Project
• Doing fieldwork and archival research for a creative nonfiction book and audio story on the human-animal relationship.
Advised by Stanford DCI fellow Melissa Dyrdahl and Prof. Jonah Willihnganz.
• Collected tens of hours of interviews and 4000+ archival documents
• Interviewed by VICE for my work with animal trainers

SELECTED COURSEWORK

Computer Science
• CS 224R Deep Reinforcement Learning (Spring 2023, A)
• CS 234 Reinforcement Learning (Winter 2023, A)
• CS 224N Natural Language Processing (Winter 2023, A)
• CS 330 Deep Multi-task and Meta Learning (Fall 2022, A)
• CS 231N Deep Learning for Computer Vision (Spring 2022, A)
• CS 229 Machine Learning (Fall 2021, A)
• CS 285 Deep Reinforcement Learning (Berkeley, self-study)
• CS 161 Design and Analysis of Algorithms (Winter 2022, A)
• CS 110 Principles of Computer Systems (Fall 2021, A)
• CS 107E Systems from the Ground Up (Winter 2021, A+)
• CS 106B Programming Abstractions in C++ (Fall 2020, A)

Mathematics
• EE 276 Information Theory (Spring 2023, A+)
• Math 115 Real Analysis (Fall 2022, A+)
• CS 228 Probabilistic Graphical Models (Winter 2022, A+)
• CS 109 Introduction to Probability (Spring 2021, A)
• Math 113 Linear Algebra and Matrix Theory (Winter 2021, A)
• Math 51 Linear Algebra and Multivariable Calculus (Fall 2020, A+)
• Integral Multivariable Calculus (2019-20, JHU Online, A+)

Psychology
• Psych 45 Introduction to Learning and Memory (Spring 2023, A+)
• Psych 30 Introduction to Perception (Fall 2022, A+)
• Psych 1 Introduction to Psychology (Spring 2022, A)
• Psych 50 Cognitive Neuroscience (Winter 2022, A+)

Writing, Literature, & Philosophy

• English 290 Advanced Fiction (Spring 2023, A)
• English 191 Intermediate Non-Fiction (Winter 2023, A+)
• English 190 Intermediate Fiction (Spring 2022, A+)
• English 127A Moby Dick & The Role of Animals in Fiction (Spring 2022, A+)
• English 92 Introductory Poetry (Fall 2021, A)
• English 91 Introductory Creative Non-Fiction (Spring 2021, A)
• Phil 2 Moral Philosophy (Spring 2021, A)

Curriculum Vitae | Last Updated 6-18-2023 Page 4