Sign in

Electrical Engineering Project

Scotts Valley, California, United States
February 08, 2018

Contact this candidate


MS. – Electrical Engineering Department

UC Santa Cruz, Ca,USA

Mobile no +18314212250


Internship at Kogence Inc.,Mountain View, CA,USA(June 2017-Sept 2017) Kogence is a modeling and a simulation startup based in Silicon Valley. Mentor was the CEO, Dr. Mukul Agrawal. On Kogence, one can search, browse, execute and latency-free interact with the graphical results on the browser of any personal device. The objective of the internship is to populate the platform with simulation examples from the field of machine learning like Neural Nets, SVMs etc. In addition, it involves contribution in the database management team for creating flat tables from the database dump via SQL and then charting useful user statistics. TECHNICAL PROFICIENCY

• Programming Languages: Python,Matlab

• Tensorflow, Keras,Theano, Numpy,SciPy

• Database Management: SQL


• Salil Kanetkar, Ayush Pathania, Vivek Venugopal, Suresh Sundaram: Offline Writer Identification Using Local Derivative Pattern. ICFHR 2016: 355-360 s


Research Internship at Georgia Institute of Technology, Atlanta, USA (May-July 2015) Guide: Dr. David Anderson(Dept. of EECS)

The internship was in the field of machine learning. The objective involved finding a novel method for object classification using NMF( Non-negative Matrix Factorization). Instead of taking the NMF of entire training images, the NMFs of segments of training images were calculated instead, and tested them on corresponding segments of test images. The testing was done using mean subtracted 2-d convolution. The results obtained showed promise and the accuracy obtained was better than the accuracy obtained by previous methods.

Research Internship at University of Central Florida, Orlando, USA (May-July 2014) Guide: Dr. George Atia (Dept of EECS)

The project involved finding selection policy for controls which would optimize the observations subject to a few parameters. The distributions of observations were defined under various controls and also the prior distributions of the parameters. The posterior distributions were calculated assuming observations as Markov chains. The task of finding the probabilities of controls was done by taking the weighted Fisher information matrices and maximizing them. The Most Likely Estimate(MLE) of the given parameters were calculated for the case for coupled parameters. PROJECTS

Speech Recognition using Convolutional Neural Networks( Sept-Dec 2017) Guide: Dr. Manfred Warmuth, Computer Science Department, UC Santa Cruz,CA The project objective was the use of Convolutional Neural Network to identify spoken words of 1 second duration each. The novel idea in this project was the conversion of sound bytes to images by taking their FFT and then treating the problem as an image recognition one. It involved using hyperparameter tuning techniques and a lot of techniques like dropoff, training suppression etc. Also attempted to use Transfer Learning on a big neural net like alexnet. Recognizing social circles in social media (March-June 2017) Guide: Dr. Lise Getoor, Computer Science Department, UC Santa Cruz, CA The project objective was to develop a ML task of identifying users’ friends to their social circles. This was done in Probabilistic Soft Learning(PSL) which is a relational learning tool. In relational learning, the inherent relationships between observations are also considered as a part of the decision process and are not considered IID as they are normally done in other machine learning tasks.

Offline writer identification( August 2015-April 2016) (Undergarduate Thesis Project) Guide: Dr Suresh Sundaram, EEE Department, IIT Guwahati In this project, a novel scheme for identifying the authorship of off-line handwritten documents was proposed through a histogram-based descriptor(LDP). LDP has previously used for face detection, however, this is to the best of our knowledge, the first work that utilizes LDP for characterizing the authorship of an writer. Classification was done by the chi-squared classifier. The developed algorithm gave state-of-the-art accuracies. ACHIEVEMENTS

Ranked among the top 0.2% students i.e AIR 1481 (among 540000 appeared students) in the internationally acclaimed IIT-Joint Entrance Examination (JEE) in 2012. Degree Institute CGPA YEAR

MS- Electrical Engineering UC Santa Cruz, USA 4.00/4.00 June,2018 B.Tech.- ECE IIT, Guwahati, India 8.04/10.0 July, 2016 RELEVANT COURSES

Probability Theory Information Theory

Advanced Machine Learning Computer Vision

Pattern recognition & machine learning

Github link-


Contact this candidate