About
I'm Akshay Manglik, an Applied AI Engineer at Scale AI on the Enterprise ML team.
I graduated from Columbia University in 2025 with a B.A. in Computer Science and Economics, and a minor in Mathematics.
I'm interested in designing adaptive systems capable of learning and self-correcting over time, with minimal human intervention.
Dereferenced is my blog about the intersection of technology, economics, and philosophy. I'm especially interested in understanding complex systems from the ground up, whether those be market systems, technological systems, or systems of ideas. Feel free to reach out to chat or with feedback on my writing!
Connect
Projects
A Letterboxd-like interface with an LLM that recommends movies based on your watchlist. Includes personal ratings, reviews, and AI-powered discovery.
Developing model and data scaling laws for test-time search and sampling methods on reasoning benchmarks, for 400M–72B parameter models.
AI-powered iOS app promoting alcohol safety via real-time blood alcohol concentration estimation. Created model to predict BAC from movement data using CNNs.
A study and practice tool for Spanish verb conjugation with per-tense spaced repetition and optional LLM feedback for explanations and corrections.
Android app that scrapes information about your California ballot and provides comprehensive, verified information about candidates' backgrounds and political stances.