SOUVIK ROY
Specialized in Machine Learning and Deep Learning
[ adcusa@r.postjobfree.com Ó 647-***-**** Waterloo, Canada
https://www.linkedin.com/in/royalsouvik/ https://www.github.com/s56roy EXPERIENCE
PROJECT ENGINEER NORTH
EAST CENTER FOR TECHNOL-
OGY APPLICATION AND REACH
June 2014 – May 2018 New Delhi, India
Worked on the development of unmanned
aerial vehicle. Responsibilities included system
designing, integration & embedded program-
ming.
Worked on ANPR - Automatic car number
plate recognition software. SVM and ANN
algorithms were used for numberplate recog-
nition and optical character recognition.
PROJECTS
Sentiment Analysis of Yelp Review Dataset
Built and explored various ML models for clas-
sification using KNN, XGBoost, Decision Tree,
Naive Bayes, Random Forest, SVM, Gradient
Boost, LSTM and MLP.
Major challenges were class imbalance, using
different word vectorizer and hyperparameter
tuning to achieve better accuracy score than
academic journels.
Ship Detection in Optical Satellite Imagery
Developed, modified and implemented pattern
recognition for ship detection and localiza-
tion in satellite images using classical machine
learning algorithms like SVM and Random For-
est for classification and for generating feature
vector using LBP, HoG and saliency map.
Cognitive Engineering Case Analysis:
Google Maps Application on Android
Examined the "Google Maps on Android" as
the problem to find out decision making ca-
pabilities and interface usage. the impact of
these human behaviour by user experience
level and usage process. Experiments were
performed using CTA & CD.
Automatic Number Plate Recognition Soft-
ware
Developed a classical machine learning SVM
model in C++ & Qt for recognizing the En-
glish characters from the segmented images
extracted from video frames. SVM model is
trained using Google tesseract OCR dataset
available in public domain.
Anomaly Detection in Autonomous vehi-
cles
The project is to find anomalies in the au-
tonomous vehicle dataset using classical ma-
chine learning algorithm and LSTM neural net-
work, thereby developing a prediction model
for the parameters and finding anomalies in
the dataset.
SUMMARY
Experience with Machine Learning skills - tensorflow, scikit-learn, NLTK, Pandas, Matplotlib and data visualization technique using Tableau, Python and R.
Developed Classical Machine Learning and Neural network Model for Sentiment analysis of yelp Food review dataset of 1 million to achieve higher results than academic papers.
Used image processing technique to generate feature vector using HoG, LBP, Color Histogram and combination of these, which was used to Classify the ships in satel- lite images. PCA was used to reduce feature size and classification score of 82% achieved with Pattern recognition methods SVM and Random Forest.
Used Python to perform ETL of dataset from Postgres & SQL database and devel- oping neural network model like LSTM.
6+ years of experience in C/C++ programming with Data structure, algorithms and SQL queries. And 3+ years of experience in machine learning. ACHIEVEMENTS
Winner of Telus Challenge at UofT Hackathon - Solved a Social problem
Participated in NASA Space App Challenge
Former Vice-president of Yantra Society
PUBLICATION
"ANPR Indian system using Surveillance Cameras" 2015 Eighth International Con- ference on Contemporary Computing (IC3), 20-22 Aug. 2015. EDUCATION / COURSES
M.Eng in System Design Engineering
University of Waterloo
May 2018 – October 2019 Waterloo, Canada
Methods and Tools for Software Engineering
Pattern Recognition
Data and Knowledge Modeling and Analysis
Intelligent Systems Design
Text Classification
B.Tech in ECE
Indraprastha University
August 2010 – July 2014 New Delhi, India
SKILLS
Programming
C C++ Python R Tableau Shell MYSQL Powershell LAT EX Jupyter
Ancillary Technologies
Linux Keras Tensorflow Scikit-learn NLTK PostgreSQL Git/Gitlab SolidWorks Qt Google Cloud Platform SPSS Data Cleaning