Srinath Srinivasan | North Carolina State University

About Me

I am a Master’s student at North Carolina State University Raleigh, USA. I am part of the TensorWorld group advised by Dr. Jiajia Li. We are working towards accelerating compressed/tensorized transformers. I am working on effective tensor contraction techniques and accelerating GEMM algorithms for this purpose.

I was previously a Software Engineer at Poshmark working on their iOS app. At Poshmark I led efforts to port several of their legacy MVC Objective-C code to the Swift/SwiftUI MVVM paradigm. We redesigned several screens from scratch including creating an efficient pagination algorithm for some screens with heavy scroll cases.

During my undergrad, I worked in the Thermodynamics and Combustion Engineering Laboratory at the Indian Institute of Technology, Madras. I worked on using Machine Learning to localize and reconstruct acoustic signals. I have also worked at the Robert Bosch Centre for Data Science and AI to predict student dropouts from MOOCs and on a project to predict ambulance timings for the Govt. of Tamil Nadu 108 ambulance service. I have also had 2 SWE internships at Navis and Renault.

Feel free to shoot me an email to discuss any opportunities/for research collaboration!

Research Interests

High Performance Computing: Systems in ML/ Accelerated ML
Machine Learning: Uncertainty-aided ML, Uncertainty quantification techniques.

Selected Publications

JASA

Machine learning aided near-field acoustic holography based on equivalent source method

Chaitanya SK, Sriraman Siddharth, Srinath Srinivasan, Srinivasan K

The Journal of the Acoustical Society of America, 2023.

Link Journal

IJA

Data-driven neural networks for source localization and reconstruction using a planar array

Sai Manikanta Kaja, Srinath Srinivasan, S. K. Chaitanya, K. Srinivasan

International Journal of Aeroacoustics, 2022.

Link Journal

IAICT

Exploring Bayesian Uncertainty Modeling for Book Genre Classification

Srinath Srinivasan, S G Shivanirudh, Sujay Sathya, T T Mirnalinee

IEEE International Conference on Industry 4.0, Artificial Intelligence, and Communications Technology (IAICT), 2022.

Link Conference

Education & Courses

North Carolina State University, Raleigh, USA (2023 - Present)
Masters in Computer Science; GPA: 4.33/4
Courses: Design & Analysis Of Algorithms, Software Engineering, Neural Networks, Efficient Tensor Processing for AI
Anna University, Chennai, India (2018 - 2022)
B.E. in Computer Science & Engineering; GPA: 8.82/10

Work Experience

North Carolina State University, Raleigh, USA
Research Assistant (Aug 2023 - Present)
- Optimizing training and inference of compressed transformer architectures for various compute/memory bound use cases
- Working on intermediate layers’ tensor contraction (GEMM) optimization through vendor/open-source libraries
Poshmark Chennai, India
Software Engineer - 1 (Aug 2022 - Jul 2023)
- Developed key infrastructure components such as screen pagination algorithms from scratch for the new SwiftUI Combine based publisher-subscriber design paradigm
- Migrated multiple legacy objective C components to SwiftUI MVVM architecture, resulting in a 60% reduction in code complexity and improved app performance
- Implemented automated testing framework for migrated components, increasing test coverage through dependency injections reducing bug resolution time
- Developed crucial components for Poshmark’s time-sensitive live streaming and auction features
Navis, Chennai, India
Associate Software Engineer - Intern (Jan 2022 - Jul 2022)
- Resolved mission critical bugs in the Navis N4 terminal operating system, increasing system stability
- Worked on an adjacent AI-enabled product to enable smart docking of ships to optimize dock space
Renault Nissan Technology and Business Centre, Chennai, India
SDE Intern (Aug 2021 - Oct 2021)
- Developed a scalable cloud-based pipeline enabling car dealerships to remotely communicate key data points to analysts
- Application reduces delay in creating analytical reports and minimizes manual entry errors through multiple validation checks