David Koplow

Personal Website

Education

Massachusetts Insititute of Technology

Degree

SB in Artificial Intellgence and Neuroscience (double major).

Sep 2020 - Feb 2024 (3.5 years)

GPA: 5.0/5

Societies

Acitvities

Select Experiences

Co-President

AI@MIT

Student Organization
Fall 2020 - Jun 2024
Managing organization logistics and communication with the 500+ associate members and dozens of general members. Developed sponsorships with industry partners such as OpenAI. Previously ran AI@MIT Labs where I managed 40 students building projects and helped resolve technical problems. Prior to that I was part of AI@MIT Labs, and built a number of projects including CV limb detection game, DALLE plant generation, and intra-building directions for campus. Designed and built our new website with Next JS aiclub.mit.edu with automatic Google Calendar integration.

Co-President

Eta Kappa Nu Honor Society

Student Organization
Fall 2020 - Jun 2024
HKN runs MIT's official EECS tutoring service and manages the largest student repository of class evaluations. Historically up to 50% of requests for tutors were not fully satisfied due to high demand and improper matching. I wrote a new algorithm that improves this match rate by 40% on historical data. Started a committee to pair students with local non-profits to solve technical problems. Established a framework for compensation and legal protections for students. The non-profits indicated on their applications these problems cost them up to $10 million dollars and affect 1000s of people.

Co-Founder

Tau Clinical

Full-time
Jun 2023 - May 2024
Worked on pre-seed startup to speed up clinical trial regulatory submission by 2 weeks using artifiial intelligence to perform automatic data standardization for FDA submissions. Solving this problem could lead to biotech companies saving $1,000,000s in lost revenue. Conducted market research and personally interviewed over one hundred stakeholders including Pharma executives and leadership at the FDA to validate the problem. Designed and ran a mock clinical trial to test out an early prototype of the software. Built a demo to show customers. Accepted into Y Combinator, Prod, MIT DHIVE, and MIT Sandbox.

Machine Learning Engineer

Pumas AI

Internship
Jun - Aug 2021
In the summer of 2021, I Interned at Pumas AI, a spinout of the MIT Julia Lab focused on using neural differential equations to improve pharmacometric and pharmacodynamic drug models to better design clinical trials. I implemented reverse mode automatic differentiation of their software to speed up Deep Non-Linear Mixed Effects models by up to 40%. This involved defining vector Jacobian products for Julia code with Zygote and compiled code for Diffractor. I also implemented a number of models into their test suite includeing Hepatitis C virus model.

Machine Learning Engineer

BrainQ Technologies

Internship
Jun - Aug 2022
BrainQ Technologies is developing the first affective treatment for strokes. Since my internship in the summer of 2022, they've completed their first round of clinical trials and shown patients using their medical device are 200% more likely to make a full recovery. While working there, I used transformers, variational autoencoders, and other natural language processing techniques, to train models able to identify changes in the brain caused by their medical device with PyTorch to inform the design of their clinical trials. To improve results, I performed feature augmentation of EEG Data for the company’s clinical trial with MNE and BrainDecode.

Software Developer

Stem Development Inc.

Consultant
Jun 2020 - Aug 2020
Wrote 140+ Physics, Biology, and Chemistry interactive JavaScript simulations that were integrated into the company’s digital textbook series that is now being used to teach every high schooler in Qatar. I streamlined their software architecture to enable quicker revisions to old and building of new simulations, ultimately saving the company's seven figure contract.

Intro

David Koplow graduated from MIT with a degree in Artificial Intelligence and Neuroscience. He co-led AI@MIT and Eta Kappa Nu, focusing on improving tutoring services and managing student organizations. As a Machine Learning Engineer intern at Pumas AI and BrainQ Technologies, he worked on automatic differentiation and applied NLP techniques to EEG data. He co-founded Tau Clinical, a startup accepted into Y Combinator, aiming to expedite clinical trial submissions using AI. His research projects include a rebuttal to a preprint on MIT's curriculum and exploring the Graph Isomorphism Problem.


(Blurb written by ChatGPT o1-preview)


Select Research and Projects

Wrote a rebuttal to ”Exploring the MIT Mathematics and EECS Curriculum Using Large Language Models” after it went viral on Twitter. The pre-print was originally co-authored by 15 researchers including 4 professors affiliated with MIT. The issues we identified culminated in the withdrawal of the paper. Our response also went viral attracting 3.4 million views and led to an article in the Chronicle of Higher Education. I was asked to speak about this to over 100 researchers and educators at a conference on LLMs in education run by MIT. List of other speakers. I also gave this talk to Tomaso Poggio’s lab and was a panelist for MIT Generative AI Week.

The Eta Kappa Nu honor society at MIT is the official tutoring service for all of MIT EECS. Historically, all pairings between tutors and tutees are done manually. Often only 30-50% of the need is met due to suboptimal pairings resulting from the stochastic nature of when sign-ups occur, the availability of tutors, and the differing levels and areas of need for students. Using historical data of student-tutor pairings and requests and LLMs, the optimization algorithm described in this paper enables us to meet over 40% more of student need. Read the paper.

Optical satellites are one of the most valuable sources of surveillance data, yet there do not seem to be any unclassified methods that verify their data in real time. We propose a pipeline that accomplishes this through the frame-by-frame verification of satellite position, image location, and temporal accuracy of the image. We also present and implement a shadow-based method of automatic single-frame image temporal verification. This model performs well on a dataset of satellite images of major cities, even when images are partially obstructed by weather anomalies and cloud cover. It can accurately approximate the time of day an image was taken within an hour, but is not yet robust enough for military applications. Read the paper.

The question of whether or not there is an efficient algorithm to determine if two graphs are isomorphic is unsolved. While there are a number of solved sub-types of graphs, no polynomial time algorithm shown to work in every case. In this paper I present an O(n⁴) polynomial time algorithm that appears to generalize to some yet unsolved subtypes of graphs. This algorithem, however, fails to identify isomorphisms in non-strongly regular graphs (commonly believed to be the most difficult subtype of graph to detect isomorphism) above 125 nodes. Thus the question of whether or not GI=P remains open. This write up of the algorithm and quantiative expirments to evaluate its performance were conducted under the supervision of Professor Virginia Williams.