Sumanth Reddy Cherupally | Software Developer & Researcher

Software Developer & AI Researcher

Passionate about software development, robotics, and deep learning. Specialized in building scalable systems and applying machine learning to solve complex problems.

About Me

I'm a software developer with expertise in building scalable systems, data frameworks, and applying machine learning techniques to solve complex problems. With a background in Electrical Engineering and a Master's in ECE focusing on Robotics and Deep Learning, I bring a unique perspective to software development challenges.

My experience spans across financial technology, autonomous vehicles, and cutting-edge research in areas like reinforcement learning, computer vision, and blockchain technology. I enjoy tackling complex problems and creating efficient, elegant solutions.

Work Experience

Barclays

August 2023 - Present

Software Development Engineer - FC Tech Frameworks | Whippany, NJ

Engineered a robust data framework that enhanced financial reporting system performance, managing over 20 Million rows of complex datasets reducing downstream data issues by 50% per cycle while adding no additional time for report generation.
Created a scalable framework named Control-Check for efficient management and validation of large datasets.
Developed and maintained web applications leveraging the Django framework, ensuring seamless performance and scalability.

Python Django Data Engineering SQL

TuSimple, Inc.

June 2022 - December 2022

Software Development Intern - Onboard Runtime Team | San Diego, CA

Ideated and implemented a GPU Tracing Tool to identify runtime spikes in CUDA API and kernel executions from onboard vehicle GPU, enhancing system performance and reducing developer hours by 75%.
Redesigned a database schema for efficient data storage and retrieval, achieving a 40% reduction in storage and a 95% improvement in query time.

C++ CUDA GPU Programming Database Design

Research Experience

Existential Robotics Lab, UCSD

January 2022 - August 2023

Prof. Nikolay Atanasov, ECE

Executed deep reinforcement learning models to end-to-end control a Vision aided car environment using policy gradient algorithms.
Led the implementation of a multi-agent simultaneous localization and mapping (SLAM) project using Distributed sparse Gaussian processes to predict TSDF values from dense point clouds using an Octree.

Deep Reinforcement Learning SLAM Computer Vision Robotics

Lab for Video and Image Analysis, IIT Hyderabad

August 2020 - May 2021

Prof. Sumohana S. Channappayya, EE

Proposed a novel depth-based classification technique to distinguish occluded from non-occluded images, achieving a recall improvement from 76% to over 95% at constant precision.
Designed advanced neural networks for object recognition using ResNet variants, enabling high-quality monocular depth estimation and efficient processing of complex 3D datasets.

Computer Vision Deep Learning ResNet Depth Estimation

Practical Networking and Blockchain Lab, IIT Hyderabad

May 2019 - April 2021

Prof. Kotaro Kataoka, CSE

Introduced a novel blockchain protocol optimized for IoT devices, focusing on scalability and resource efficiency.
Developed communication protocols using Golang with multi-threading and channels for concurrency control.
Published two papers in the Journal of Network and System Management and ICOIN 2021.

Blockchain Golang IoT Distributed Systems

Education

University of California, San Diego

MS in Electrical and Computer Engineering (Robotics and Deep Learning)

September 2021 - July 2023

Key Coursework:

Deep Learning Reinforcement Learning Advanced Computer Vision Linear Algebra Motion Planning and Control

Indian Institute of Technology Hyderabad

B.Tech in Electrical Engineering

July 2017 - June 2021

Projects

Parallel Computation

Optimized matrix multiplication performance by: (1) 1100x on NVIDIA GPUs using OpenMP and CUDA programming to optimize occupancy along with blocking and tiling techniques. (2) 7x on CPU using cache optimization techniques such as blocking and packing, as well as instruction-level parallelism with intrinsics.

CUDA OpenMP C++ MPI

Neural Radiance Fields (NeRF)

Implemented and studied the canonical NeRF and its multiple variants like Mip-NeRF and Pixel-NeRF. Tested and presented multiple synthetic 3D views for novel view synthesis applications.

PyTorch Computer Vision 3D Rendering Deep Learning

LSDI: Blockchain for IoT

Designed and implemented a lightweight and scalable DAG-based distributed ledger for verifying IoT data integrity, with repositories available on GitHub.

Golang Blockchain Distributed Systems IoT

Gateway Repository Node Repository

Publications

Lightweight and Scalable DAG based distributed ledger for verifying IoT data integrity

S. R. Cherupally, S. Boga, P. Podili and K. Kataoka

2021 International Conference on Information Networking (ICOIN), 2021

View Publication

Inter-domain prefix and route validation using fast and scalable DAG based distributed ledger for secure BGP routing

Prashanth Podili, S. R. Cherupally, Srinivas Boga, Kotaro Kataoka

Journal of Network and Systems Management, Special Issue on Blockchains and Distributed Ledgers in Network and Service Management

View Publication