Eklavya
Sehrawat
Program Manager • AI/ML Engineer • Full-Stack Developer
About Me
The Journey
I'm an experienced Program Manager at Amazon with deep expertise in ML/LLM systems. I excel at transforming complex AI research into scalable, production-ready systems that people actually love to use.
From building multi-agent AI frameworks on AWS to modernizing legacy BI infrastructure with React, I bridge the gap between cutting-edge research and practical solutions that scale to thousands of users.
Current Focus
Agentic AI & Large Action Models
AI & ML
LLMs, Transformers, RAG Architectures
Full-Stack
React, Next.js, Modern Web Apps
Hardware
PCB Design, Embedded Systems
Location
Austin, Texas
Experience
Feb 2022 – PresentSenior Program Manager
Supply Chain AI/ML Team • Austin, TX
- ▸ Deployed an AI-powered demand forecasting system serving 321+ delivery stations (50.2M packages/week) across US/Canada, achieving $9.7M YoY cost savings through multi-agent AI orchestration frameworks leveraging AWS Step Functions, Lambda, and SageMaker.
- ▸ Engineered hybrid machine learning pipelines combining foundation models (Gluon, Prophet) with fine-tuned transformers on 18+ months of historical demand data, integrated with RAG architecture for real-time signal ingestion. Delivered 6% forecast accuracy improvement.
- ▸ Built production-grade LLM application using AWS Bedrock (Claude 3 Sonnet) with semantic search powered by PostgreSQL pgvector for 50M+ embedding vectors. Scaled to 2,000+ daily active users including C-Suite executives.
- ▸ Modernized legacy BI infrastructure by building React-based analytics websites with Shadcn UI components, integrating ML-driven insights and PostgreSQL backends. Reduced dashboard load times by 40% and infrastructure costs by $12K/month.
Jan 2020 – Feb 2022Program Manager
Amazon Logistics • Toronto, CA
- ▸ Managed team of 7 technical specialists to scale learning and development platforms across Amazon's logistics network (22K employees, 12M+ customers served). Implemented data-driven training systems with automated skill tracking.
- ▸ Spearheaded COVID-19 response initiative, independently building full-stack web application for real-time employee location tracking across warehouse facilities. Engineered contact tracing system with shift-based cohort isolation logic. Deployed in <3 weeks.
Projects
Hawaii Expansion for Amazon
Launched the first Amazon delivery station in Honolulu in 13 months. Built operational planning models and route optimization algorithms that achieved $4.3M annual cost savings by reducing third-party reliance while improving delivery speed by 38%.
AI-Powered Supply Chain Forecasting
Built a demand forecasting system leveraging multi-agent AI orchestration, serving over 321 delivery stations. Combines foundation models with RAG architecture for real-time signal integration.
Enterprise LLM Analytics Platform
Production-grade LLM application using AWS Bedrock with Claude 4.5 Sonnet and semantic search powered by PostgreSQL pgvector. Full-stack implementation serving C-Suite executives with 300+ daily active users.
Tech Stack
Frameworks & Languages
Tools & Platforms
Education
McMaster University
Bachelor of Commerce (Dropped Out)
Technical Focus: Advanced coursework in software development (iOS/Swift, Python), data structures & algorithms, and statistical modeling. Built strong foundation in business analytics and technical problem-solving.