Amartya Ranganathan
I build secure, robust infrastructure that enables machine learning systems to operate reliably at real-world scale.
Currently, I'm a software engineer on the Public Key Infrastructure team at GEICO Insurance, where I work on the security and trust primitives that all the company's domains and APIs depend on. Before that, I built cloud-native infrastructure on the application registry team at Amazon Web Services, developing systems used globally. I have also seen the power of generative and quantitative methods applied in industry while building AI-driven educational products at LearnWith.AI and applying rigorous economic analysis at Cornerstone Research to influence consequential litigation outcomes. Across these roles, I have developed a strong foundation in cloud computing, infrastructure engineering, and applied mathematics, and I aspire to build the foundational systems that allow transformative ML models to be deployed efficiently, securely, affordably, and sustainably.
I hold a Bachelor of Arts in Economics and Bachelor of Science in Applied Mathematics from the University of Southern California. Fight on!
Jane Street Market Classification
We explored the power of quantitative modeling to generate profitable predictions in financial markets. After estimating each asset’s drift over a specific time horizon using Geometric Brownian Motion, a Mixture Density Network made probabilistic trading decisions based on the conditional distribution of 130 features.
ReadAct
ReadAct is a browser extension that uses natural language processing to highlight pivotal issues, candidate positions, and political bias in today’s bipartisan landscape. Leveraging cloud computing, APIs, and machine learning, our team delivered an intuitive and visually polished product on the Chrome Web Store.
Vibecheck
How does music capture the emotions of our conversations? Vibecheck uses speech-to-text and sentiment analysis to capture feelings (like "happy" and "chill") and relate them to music features like beat and danceability.
Allai
Natural language processing revolves around vocabulary, sentiment, and syntax. Allai targets inclusive language by leveraging these attributes to process users' input and provide feedback to make the messaging empathetic and inclusive.