Meta Superintelligence Labs (MSL)

Meta AI growth and discovery (LLM and VLM model scaling, feature development and AI video generation model capability scaling)

Growth: notifications, messaging, sharing infra, upsells and feed ranking for AI products (Meta AI Vibes, continue conversation, AI feed)

I build data pipelines, APIs and UI tools for regulatory data collection and supporting compliance for cross product data use.

Full stack vehicle diagnostics software development, in React, TypeScript and Python3. Developed and deployed critical full stack software components, including consumer-facing UX and firmware upgrades.

Worked on building graph neural networks to solve the node selection problem in MIP solvers.

Project at ROBITA Lab, IIIT Allahabad, in association with Mercedes Benz R&D India.

Working on augmenting existing Simultaneous Localisation and Mapping (SLAM) approaches (RTAB-MAP, S-PTAM) using Semantic Segmentation methods based on visual traffic data. Tasks included:

• Collection of data

• Developing a new dataset with multiple classes absent in SOTA datasets

• Transfer learning on Mask R-CNN based on new data

• Integration of Mask R-CNN into SLAM architectures