Electrical Engineering Software Engineer

Kunal Ashok Mishra

San Jose, CA ac55df@r.postjobfree.com 669-***-****

https://www.linkedin.com/in/kunal-mishra-258a82122/ https://github.com/djkunj2010

Summary:

Master’s in Electrical Engineering, a data-scientist aspirant, focused & passionate to work for service industry such as SaveSoft, Inc.

Education:

M.S in Electrical Engineering, San Jose State University, San Jose, CA Spring 2017

B.E in Electronics Engineering, Visvesvaraya technical university, Belgavi, India June 2013

Technical /Applied Skills:

Operating system

Windows, Linux,

Programming languages

Python, R, C++,

Concepts

Data Science, Machine Learning, Deep Learning, Artificial Intelligence, CNN (Convolutional neural network), RNN (Recurrent neural network), NLP (Natural language processing), Descriptive Statistics, Logistic Regression, Linear Regression, A/B testing, CPU and GPU Architecture, SQL, Data-structures, Algorithm, FPGA, ASIC-Design,

Libraries and Tools

Keras, Tensorflow, Tableau, Numpy, Pandas, Scikit-learn, Scipy, Matplotlib, R Studio, Anaconda, Spyder, Jupyter-Notebook, Eclipse (C Matlab,

Professional Expertise:

Intern at SSRLABS, San Jose, CA July 2017 - Current

Created a communication path between RISC-V and CNN core, and tested on spike simulator.

Sports official at San Jose state university, San Jose, CA Jan 2017 - May 2017

Manage sports team and events, such as Soccer, Volleyball, Cricket, Basketball, etc., and provide conflict solution to players.

Associate software engineer at Reliance group of industry, Mumbai, India. Sep 2013 - Apr 2014

Selected to be a part of team network team, to implement the network infrastructure for HNH hospital, the largest hospital in Mumbai.

Serving better quality of modernized technology to the patient and providing safety for all the people present in the hospital.

Academic Projects:

Customer Churn Problem of a bank Independent

Created an artificial neural network model to solve a fictional customer churn problem of a bank.

A model was trained by using a sample data of bank, to predict the number of customers terminating an account, with python, Keras libraries.

The model achieved an accuracy of 83.6%, and further efficiency was improved by using K-fold cross-validation, dropouts, GridSearchCV.

Stock Price Prediction Independent

Modeled a recurrent neural network to predict Google stock price for the financial year of 2017.

The model was created by using LSTM, to train 5 years of Google stock price, with Python, Keras libraries and Matplotlib for visualization.

Results for the month of January was predicted, output was very close to the actual real-time behavior of the stock. Further, it was improved by using additional LSTM layers, more data of the stock price and GridSearchCV.

BI for public transit SJSU Spring 2016

Developed a business intelligence solution for Valley Transport Authority.

Where route no. 22 and 522 were compared (both the bus runs on the same route), by using the data set provided by VTA website.

KPIs, BSC was created for VTA, visualization of the routes was shown using Tableau and an optimize solution was achieved by our group.

Implementation of Chatbot (NLP) Ongoing

Designing a Chatbot using a Cornell movie dialog corpus dataset, and integrating the dataset with RNN design algorithm.

The project is designed in 4 parts, data-preprocessing, building seq2seq model, training and testing the seq2seq model.

Image Recognition (CNN) Independent

Designed a convolutional neural network model to predict the image whether the image is ‘dog’ or ‘cat’.

Image data in the model was convoluted, max pooled, flatten and connected to form input for the classic neural network using Keras.

The accuracy of 86.26% was achieved for training model and 81.25% for the test model, further model was improved by using Adam-Optimizer, more training & test dataset, increasing the number of hidden layers and dropouts.

SIGN language recognizer from number 0 to 5 Independent

Shaped a Deep Neural Network model, using convolutional neural network technique to recognize sign language.

A SIGNS dataset was collected to train the authentic CNN model, with Python, Numpy, and Tensorflow libraries, the model was created.

Train and test accuracy was 94.07%, 78.33% respectively, and performance was enhanced by hyperparameter tuning and L2 regularization.

Creating Jazz Music using RNN Independent

The Composition of Jazz music by an RNN model using LSTM layer technique.

A corpus of Jazz music was trained to the LSTM model, to achieve an improvisation of Jazz music, created by a deep learning.

Relevant Coursework:

Business intelligence, CMPE 274 at San Jose State University.

Deep learning A-Z™: Hands-On Artificial Neural Networks by Kirill Eremenko and Hadelin de Ponteves at Udemy.com.

Contact this candidate