Chandru Pokala

About Chandru Pokala

I build engineering organizations that turn complexity into clarity — systems that scale not just in compute, but in trust, collaboration, and purpose.At AWS AI (Annapurna Labs), I lead the teams behind Neuron Containers, Tools, and AI Platforms — the exascale ML infrastructure powering Bedrock, SageMaker, and Alexa. Our platforms orchestrate hundreds of thousands of Trainium and Inferentia accelerators to deliver billions of inferences with 99.95% reliability across global clusters.But scale alone isn’t the story. What drives me is building the human architecture behind these systems — helping engineers grow into leaders, creating clarity in chaos, and cultivating the kind of trust that makes innovation sustainable.Over the past two decades at AWS and Microsoft, I’ve guided multiple 0→1 and hyperscale transformations — from launching AWS ECR Public and Finch (open-source runtime now used by 200K+ developers) to re-architecting Microsoft Intune into a global service managing 500M+ devices. Each chapter has reaffirmed a simple truth: technology evolves fastest when people feel seen, safe, and inspired.My leadership philosophy:Clarity creates velocity. Trust sustains scale. Culture multiplies impact.

Experience

1. Head of Ai Ml Platform

   Amazon Web Services (AWS)

   Jan 2023 — Present · US

   I lead the Containers, Tools, and AI Platform Services teams behind AWS Neuron — the exa-scale ML infrastructure powering Bedrock, SageMaker, and Alexa. Our platform orchestrates hundreds of thousands of Trainium and Inferentia accelerators to run thousands of distributed training jobs and serve billions of inference requests with 99.95 % uptime across global clusters.I built these organizations from the ground up — growing from 3 to 40+ engineers across multiple domains (ML Infra, Containers, AI Services, and Tools) — and shaped them into one of AWS AI’s most trusted platform groups.Highlights & ImpactBuilt an exa-scale, hyperscale cluster platform based on EKS for model training and serving — powering billions of inferences and thousands of training jobs across Neuron, Alexa, and other AWS AI teams. Designed for fairness, efficient scheduling, resiliency, observability, and cost-aware orchestration, forming the backbone of AWS’s Trainium/Inferentia ecosystem.Built and scaled the unified ML Platform for GenAI workloads, enabling same-day onboarding of new open-source models (e.g, Llama 3 8B) and reducing training cost ~20 % for multiple Fortune 100 customers.Launched Predictive Health and Self-Healing automation, using ML-based anomaly detection to prevent downtime and save $100 M+ annually in wasted compute.Drove container strategy across EKS, ECS, and Neuron, introducing device scheduling, DLCs, and node-problem recovery that raised reliability to 99.99 %.Created the Neuron Tools Suite (Profiler 2.0, Debugger, Monitor, Pre-Check), cutting average debug time by >50 % and boosting developer velocity.Scaled from 3 → 40+ engineers and promoted 25 % into lead or manager roles, building a strong leadership bench and cross-AWS mentorship program.Championed a people-first engineering culture, fostering clarity, trust, and collaboration across distributed teams.

Education

• New Jersey Institute of Technology

  Master's degree, Computer Science

• Bharathidasan University

  Bachelor of Science - BS, Computer Science and Engineering