Saravanan Govindarajan

Machine Learning Engineer at Meta · M.S. Electrical Engineering, Columbia University

I am a Machine Learning Engineer at Meta working on retrieval, personalization, and evaluation for large-scale AI systems. I develop embedding-based methods, approximate nearest-neighbor retrieval, and scalable evaluation systems that operate across hundreds of millions of entities.

Previously, at Meta Reality Labs, I developed learning methods for wearable computing using vision, audio, and motion signals. My work covered human activity understanding, model generalization, and resource-efficient deployment on Meta smart glasses.

I am interested in research engineering at the intersection of representation learning, multimodal understanding, retrieval, and efficient large-scale AI.

saravanang9999 [at] gmail.com  /  GitHub  /  LinkedIn  /  Google Scholar

Saravanan Govindarajan hiking in the Dolomites

Selected Research & Engineering Work

Large-Scale Retrieval and Personalization

I develop embedding-based retrieval and personalized benchmarking systems that integrate image, text, and graph data. My work spans representation learning, approximate nearest-neighbor search, similarity modeling, and scalable evaluation, supporting retrieval and benchmarking across hundreds of millions of entities.

Wearable Intelligence and On-Device Learning

At Meta Reality Labs, I developed learning algorithms for wearable sensing using vision, audio, motion, and language signals. My research explored cross-modal alignment, efficient neural architectures, egocentric activity recognition, health-related sensing, and model generalization under real-world conditions, with a focus on reliable deployment on Meta smart glasses.

Machine Learning for Healthcare

My research includes early sepsis prediction from electronic health records and human activity understanding from wearable sensors. I developed models that learn from noisy, real-world data, with a focus on reducing false alarms and improving generalization across users, environments, and clinical settings.

Research

Awards & Recognition