Nov 2021 – Mar 2025 · 3 yrs 4 mos · Arlington, Virginia, United States

• Amazon AI/ML Services, SageMaker ML Inference, SageMaker ML Training
• I was a tech lead on various projects involving health check features, model monitor features, data capture, multi model endpoints, new region build-up for services etc. And In process, Improving the availability and fault tolerance of the machine learning system, observability of the ML models hosting, supporting and releasing features and services impacting several $xxxM in revenue.
• Checkout
• Inference, https://docs.aws.amazon.com/sagemaker/latest/dg/how-it-works-deployment.html
• Training, https://docs.aws.amazon.com/sagemaker/latest/dg/how-it-works-training.html

Jun 2021 – Oct 2021 · 4 mos · Arlington, Virginia, United States

• I was the single thread owner of the Prime Video title deduplication project. Duplicate titles negatively affected customer UX and purchase. I handled design, dev and QA, delivering in 3 months (before planned ETA). Impact - 1xx,xxx duplicate titles merged.

Oct 2017 – Aug 2019 · 1 yr 10 mos

• I worked on scaling an object storage system quadrupling it to XX PetaBytes in 18 months . Learnt a great deal about scaling, reliability, and maintain‑ability of distributed systems.
• Responsibility and Projects –Lead engineer on projects involving Object storage life cycle management, SLA dependent cluster federation,traffic routing, performance analysis, integration of PAAS services, customer data migration ,and KTLO duties.

Dec 2015 – Oct 2017 · 1 yr 10 mos

• Technical Leadfor features such garbage collection and physical erase in the Multimedia Management Layer (MML) for the Next Gen SanDiskAutomotive/Industrial NAND flash products (MLC and TLC). These are highly reliable products for industrial/automative usage.
• Project ownership, Planning and Mentorship: Coordinated with multiple stakeholders to deliver product on time, mentored new engineers,ramp up and knowledge transfer.

Jul 2011 – Dec 2015 · 4 yrs 5 mos

• Video Codec Development: I coded video codec sub‑modules‑ motion estimation, transformation and quantization block of H.265, H.264 and VP8; ported to GPU. Performance improvement of 2%‑5%for encoder profiles.
• HW Algorithms for Image Compressor: I researched and implemented fixed point FFT techniques in JPEG. This saved 3% – 5%in gate count, translating to $xxx,xxx saved over multiple product launches.
• Leading a team of two engineers on projects involving video compression modeling and improvement.