CV

The pdf version of cv can be found here

Education

• B.Tech in Electrical Engineering, NIT Srinagar, 2015
• M.S. in Electrical Engineering, Arizona State University, 2018
• Ph.D in Electrical Engineering, Arizona State University, 2022

Work experience

• Present: Senior Machine Learning Engineer, Qualcomm, San Diego, CA
  • Duties included: Developing memory efficient deep learning models using neural network architectures such as, Long Short Term Memory (LSTM), Recurrent Neural Networks (RNNs), Convolutional Neural Networks (CNNs) and Transformers for multiple applications including Speech Separation and Speech Enhancement for products in VoiceUI and infotainment. Conducting experiments to tune relevant hyperparameters such as learning rate, chunk size and weight decay to make the model real time realizable. Convert trained models to ONNX format for on-chip implementations for multiple downstream products. Documenting research findings in manuscripts submitted to different publishing venues (NeurIPS, ICASSP, Interspeech) and developing live demonstrations to effectively communicate the application.
  • Supervisor: Asif Mohammad
• Summer 2020: Student Computing Intern, Lawrence Livermore National Labs, CA
  • Duties included: Developing a structured meta-learning paradigm for effective domain adaptation and multi-task learning and showcased the effectiveness of the method using datasets such as iMaterialist, DeepFashion, iNaturalist, OfficeHome, PACS, CUB-200 and DomainNet. Performing a study using knee X-ray dataset to derive the meta information such as age, sex, gender and BMI information. This study was performed to understand whether such medical image data, preserves the privacy of the patients or if it is relatively easy to extract meta-information clues using machine learning. To this end, individual Efficient Net based deep learning models were developed in Pytorch for each meta information criterion. Developing encryption strategies using data manipulation techniques such as, mixup to perform patient data de-identification for safe data sharing within healthcare organizations. Developed privacy preserving machine learning models that were trained on conventional medical image data with a constraint such that, the trained models can also perform with high accuracy on encrypted data for machine learning based diagnostic applications.
  • Supervisor: Jayaraman J. Thiagarajan
• Summer 2018: Data Scientist Intern, Prime Solution Group, AZ
  • Duties included: Developing a python script to pre-process multivariate weather data to study the seasonal trends and make the data suitable for predictive modelling. Developing a deep learning based predictive model using recurrent neural networks (RNNs) to perform irradiance prediction using weather attributes including cloud type, temperature, relative humidity, etc. Developing a deep learning based joint predictive model using Long Short Term Memory (LSTM) Neural Network for photovoltaic power forecasting based on multivariate weather data for the Arizona region.
  • Supervisor: John Cadigan

Skills

• Programming Languages: Python (Numpy, Scipy, Pandas, OpenCV), Matlab, C, C++
• Tools and Libraries: Pytorch, Keras, Tensorflow, Git, Audition, SPSS, LateX, Microsoft Visio

Publications

Shading prediction, fault detection, and consensus estimation for solar array control

Sameeksha Katoch, G. Muniraju, S. Rao, A. Spanias, P. Turaga, C. Tepedelenlioglu, M. Banavar, and D. Srinivasan (2018) " Shading prediction, fault detection, and consensus estimation for solar array control." . 2018 IEEE Industrial Cyber-Physical Systems (ICPS) 15-18 May 2018

Fast Non-Linear Methods for Dynamic Texture Prediction

Sameeksha Katoch, P. Turaga, A. Spanias, C. Tepedelenlioglu,(2018) " Fast Non-Linear Methods for Dynamic Texture Prediction." . 25th IEEE International Conference on Image Processing (ICIP) 7-10 Oct. 2018

Audio Source Separation via Multi-Scale Learning with Dilated Dense U-Nets

Sameeksha Katoch, V.S. Narayanaswamy, J.J. Thiagarajan, H. Song, A. Spanias (2019), " Audio Source Separation via Multi-Scale Learning with Dilated Dense U-Nets."

Adaptive Video Subsampling For Energy-Efficient Object Detection

Sameeksha Katoch, D. Mohan, S. Jayasuriya, P. Turaga, A. Spanias, (2019), " Adaptive Video Subsampling For Energy-Efficient Object Detection." . 53rd Asilomar Conference on Signals, Systems, and Computers 3-6 Nov 2019

Invenio: Discovering Hidden Relationships Between Tasks/Domains Using Structured Meta Learning

Sameeksha Katoch, K. Thopalli, J.J. Thiagarajan, P. Turaga, A. Spanias (2019) " Invenio: Discovering Hidden Relationships Between Tasks/Domains Using Structured Meta Learning."

DDxNet: A Multi-Speciality DiagnosticModel for ECG and EEG

J.J. Thiagarajan, D. Rajan, Sameeksha Katoch, A. Spanias (2020) " DDxNet: A Multi-Speciality Diagnostic Model for ECG and EEG," in Nature Scientific Reports.

Teaching

• Teaching assistant
  1. EEE407/591 - Digital Signal Processing
  2. EEE334 - Circuits II

Awards

• IEEE Al Gross Award, IEEE, April 6 2019