Michael Tesfaye

Michael Tesfaye

Thinking about systems, ML, and markets

I'm a CS and Finance/Statistics student at Penn (M&T), graduating May 2026. I'm interested in low-level systems, machine learning, and their applications in trading and markets.

2025

I was a quantitative trading intern at Jane Street in New York. Conducted two research studies for my trading desks (Fixed Income and Commodities). Fixed Income: modeled adversity in bond request-for-quote auctions. Commodities: developed hedging models for natural gas basis futures; wrote a production notebook run daily to generate historical hedged return data for the desk. Developed proprietary trading systems to algorithmically trade different exchange scenarios.

2023–2024

I was in the Computer Science Research Mentorship Program at Google (remote). Worked closely with mentor Dr. Carlos Esteves on multimodal deep learning (e.g. CNNs, GPTs).

2023

I was a Software Development Engineering intern at Amazon Web Services in Seattle. Designed and developed a Java-based data engine pipeline exporting construction management data from Procore to Amazon S3 for over 10k projects every 12 hours (SNS, SQS, Fargate).

2022–2026

Jerome Fisher Program in Management and Technology, University of Pennsylvania. B.S.E. in Computer Science, minor in Mathematics (SEAS); B.S. in Economics, dual concentration in Finance and Statistics (Wharton). Graduating May 2026.

teaching

CIS 1600 — Discrete Math for Computer Science TA helped students learn the material by grading assignments, writing some homework problems, running recitations and office hours, and working with the professors to support the course.

publications

Factor Dimensionality and the Bias–Variance Tradeoff in Diffusion Portfolio Models
ICLR 2026 Workshop on Time Series in the Age of Large Models
Avi Bagchi†, Michael Tesfaye†, Om Shastri

In progress

Diffusion Factor in Portfolio Optimization
With Prof. Minshou Chen

Smaller papers

On the PAC Learnability of Distortion-Free Language Model Watermarks
CIS 6250 Final Project, 2025. Advised by Professor Michael Kearns.
Avi Bagchi†, Michael Tesfaye†

technical projects

See more on GitHub.

Kalshi Market Making a senior design project implementing semi-systematic market making and taking for Kalshi and Polymarket live sports.

CIS 5050: Software Systems a distributed cloud platform in C++ with stateless frontend web servers and a partitioned, replicated backend key-value store supporting PUT, GET, CPUT, and DELETE operations. The system includes fault tolerance through replication, checkpointing, logging, and coordinator-driven failover. The frontend is a multithreaded HTTP/1.1 server with cookies, persistent connections, and load balancing, supporting applications such as webmail and cloud storage.

CIS 5480: Operating Systems a UNIX-like operating system in C running within a single kernel process using pthreads. Implemented a FAT-based virtual file system, FCFS priority scheduling, and a set of core system calls. Developed a shell with support for piping, job control, and file system interaction.

selected coursework

Systems & Architecture
Theory & Algorithms
Machine Learning & Optimization
Quantitative Finance
Capstone & Applied Work