I craft precision models, blending efficiency with optimized FLOPS.
I am a seasoned engineer with ~8 years of experience in building solutions. I specialize in large-scale model (8B to 405B) enablement and optimization on Habana Gaudi 2 & 3 accelerators. Expertise in distributed training parallelism strategies, device and host compute/memory profiling.
Contributed to Linux Foundation’s securefederatedai/openfl
, extending federated learning with JAX/FLAX and added more features.
Additionally, I’ve developed an E2E deep learning based Intel’s Automated Vision Checkout edge AI solutions, optimized model inference on iGPU devices and deployed at various retailer site in India. Spearheaded cost-effective event driven data pipelines on AWS, delivering a significant operational cost reduction and faster time to data publication supporting ~40m API hits monthly.
MSc. Artificial Intelligence
Liverpool John Moores University
7th Summer School, Machine learning
IIIT Hyderabad
PGD in Machine Learning & AI
IIIT Bangalore
B.E. in Electronics & Communication
Visvesvaraya Technological University