Engineering theIntelligence of Tomorrow!

Leading development of production ML infrastructure for Claude models, focusing on inference optimization and safety systems.

• -Reduced inference latency by 40% through custom CUDA kernels
• -Built real-time model monitoring system tracking 50M+ daily requests
• -Architected multi-region deployment pipeline with 99.99% uptime
• -Optimized model serving costs while maintaining sub-100ms latency

Developed core infrastructure for training and serving large language models across Meta's family of apps.

• -Scaled training infrastructure to support 100B+ parameter models
• -Implemented distributed training optimizations saving $2M/year
• -Led migration of 50+ production models to new serving platform
• -Established ML infrastructure best practices adopted by 200+ engineers

Built machine learning models and infrastructure for an early-stage AI company. Contributed to core product ML features and established ML best practices.

• -Built recommendation system serving 50M+ daily predictions
• -Developed feature store processing 10B+ features daily
• -Reduced model serving latency by 65% through optimization
• -Created ML evaluation framework and testing infrastructure