Poshak Pathak
Computer science and math student at the University of Louisiana Monroe (graduating December 2026) who builds things with data. Undergraduate research assistant working on a GAN-based defense against data poisoning in federated learning, founder of the ULM Chess Club, and writer about baseball and cricket with real data. Looking for a full-time data or ML role in Dallas–Fort Worth.
Writing
- Can you beat the S&P 500 with $200 a month? Three investing strategies backtested since 2010.
- What is a stolen strike worth now? Catcher framing in the first season of ABS challenges.
- The Challenge Problem How MLB teams spent their ABS challenges in 2026, and the 36 wins they left on the table.
- Ohtani’s impact in year 2 with the Dodgers
Research
- Spending a Scarce Option Under Perceptual Noise: Optimal Challenge Policy for Major League Baseball’s Automated Ball–Strike System (alphaXiv, 2026)
- Immersion in second language does not always attenuate and attrite first language (International Journal of Bilingualism, 2026)
- Trilingual parallel processing: Do the dominant languages grab all the attention? (Bilingualism: Language and Cognition, 2025)
- GAN defense against data poisoning in federated learning (ITNG 2025)
- Bilingual Stroop effect in Nepali–English bilinguals (Nepalese Linguistics, 2022)
- Second-language instruction and L1 abilities (Journal of Cultural Cognitive Science, 2021)
- All publications on Google Scholar
Projects
- GAN Defense: real-time detection of poisoned client updates in federated learning.
- Chess Engine: neural network evaluation with alpha-beta search, tested over 1,562 games against Stockfish 14.
- BayesSurgeon: statistical analysis of post-surgical breast cancer data (Hacklytics 2025, 2nd place).
- Pictionary Plunge: a 15-category sketch classifier tuned with Bayesian optimization.
- Martingale Simulator and Imposter Game
Contact
Résumé (PDF) · poshakcodes@gmail.com · LinkedIn · X · GitHub