Megha Agarwal
San Francisco, California, United States
Summary
🤩
RockstarMegha Agarwal is a Member of Technical Staff with a decade of experience building and scaling ML systems, currently focused on LLM inference at Anthropic after leading generative AI and Mosaic research efforts at Databricks. She has a strong track record deploying low-latency, high-throughput ML services in production—from serving 800K+ requests per minute at CRED to on-device DNN optimizations at Samsung—and bridges research and engineering fluently. Her academic work at UC San Diego produced a novel framework for auditing image classifiers using counterfactual latent representations, reflecting a knack for interpretable, data-efficient solutions. Comfortable across cloud-scale pipelines (Spark, Kafka, Redshift), model optimization in C++, and causal/counterfactual analysis for business impact, she operates at the intersection of systems, ML research, and product metrics. Based in San Francisco, she combines deep technical execution with experience teaching and mentoring graduate students, bringing both rigor and practicality to production AI.
10 years of coding experience
Github Skills (32)
pytorch10
runtimes10
tensorrt10
llama10
inference10
tpu10
amd10
llm10
nvidia10
openai10
natural-language-processing10
foundation-models10
cuda10
transformer10
gpt10
Programming languages (2)
Github contributions (5)
megha95/vllm
Nov 2023 - Sep 2024
A high-throughput and memory-efficient inference and serving engine for LLMs
Contributions:81 pushes, 8 branches in 10 months
megha95/llm-foundry
Feb 2024 - Feb 2024
LLM training code for MosaicML foundation models
Contributions:4 pushes, 1 branch in 1 day
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.