Designed and built infrastructure to support the next generation of supply chain automation- Worked with a wide range of technologies including Kafka, Postgres, Bigquery, Vertex AI, Kubernetes, GCP- Built an optimized simulation framework capable of running billions of simulations per hour- Built the backend infra to support app used to collect inventory data and predict future order volumes- Languages: C++, Go, Python, SQL, Typescript

- Building, maintaining and optimizing end-to-end infrastructure to support automatic ad creation using computer vision and deep learning.- Redesigned and re wrote the infrastructure to utilize “Seam Carving” to generate ads of different sizes when provided images of a given size- Redesigned the core infrastructure of Arish, utilizing load shedding techniques, request aggregation, server throttling, etc- Led technical sessions and helped optimize distributed pipelines and queries, sometimes achieving optimizations of >200%- Collaborated with client teams to onboard new models and algorithms and design end to end data pipelines.

- Led the SLAM team of 6 software engineers at Nod in developing and implementing a SLAM algorithm on a severely constrained hardware platform.- Optimized the algorithm to achieve accuracy to within 5 mm, running real time on an Intel Myriad 2 Vision Processing Unit, leveraging the vision modules on the processor to enable parallelization of computations.- Optimizations involved learning features that tracked well and prioritizing those features over others.- Implemented stereo online calibration

- Developed computer vision algorithms for generating HD maps from image and LIDAR data.- Devised an innovative method to track lanes detected using machine learning across consecutive images and generate road lanes for HD mapping.

- Responsible for the firmware porting of the Gateway, Repeater, and Monitor devices in a wireless sensor network system to newer hardware.

- Worked as the core member of the Digital Imaging Team, responsible for implementing image processing algorithms on the Samsung Smart Camera.- Analyzed state-of-art image processing algorithms and techniques for implementation in support software for Samsung Smart Camera.- Developed and maintained support software applications for the Samsung Smart Camera of desktop, cloud and androidImplemented Image processing algorithms on the Microsoft Azure Cloud platform.Developed a mobile front-end and the corresponding backend for the application of image and video processing algorithms on the cloud- Developed the pipeline and backend for a web-based solution to update the firmware on the Samsung Smart Cameras.- Developed an application frontend for encoding and decoding videos the High-Efficiency Video Codec(HEVC) format