About

Let me introduce myself.

I am an EDG engineer at MathWorks. My research interests lie in Deep Learning, Data Science and Simulations.

Fullname: Meghendra Singh
Phone: +1(XXX) XXX-XXXX
Email: meghs@vt.edu

Profile

I have four years of professional work experience as a software developer and a strong background in computer science, software engineering and data science. I am experienced with machine learning and deep learning frameworks like: TensorFlow, Keras and Scikit-learn. I am also a Microsoft certified specialist, programming in HTML5 with JavaScript and CSS3. My current research focuses on improving reliability of large-scale interaction-based simulators using reinforcement learning.

Skills

I am proficient in full stack web and mobile development, designing web services, applied research, rapid prototyping and data analytics.

90%
Java
85%
Python
75%
C & C++
95%
HTML5
75%
CSS3
70%
JavaScript
85%
R

My Credentials Download my CV

Resume

Here are my credentials.

Work Experience

Grad. Research Assistant

August 2016 - Present

Network Dynamics and Simulation Science Laboratory, Blacksburg, VA

Working in the Social Pathways project, on data-driven agent behavior modeling algorithms for large-scale epidemic simulators. These simulators enable federal agencies like NIH and DoD to make informed and scientifically-backed public health policies.

Software Developer

May 2012 - June 2016

TCS Innovation Labs, Pune, India

Developed information systems to facilitate data collection and curation. Design and development of workflows and dashboards for simulation life cycle management. Performed data analysis and modeling tasks on collected and simulation generated data. Wrote design documents, technical reports and SDLC artifacts.

Team Lead

January 2012 - May 2012

Tata Consultancy Services, Trivandrum, India

Conducted initial research and proof of concept for projects along with software and architecture development. Lead a team of five, in a project involving design, development, testing and maintenance for an insurance services organization.

Education

Master's of Science in Computer Science

August 2016 - Present

Virginia Tech, Blacksburg, VA

GPA: 3.9/4.0
Coursework: Data Analytics, Theory of Algorithms, Computer Vision, Software Engineering, Statistics in Research, Research Methods in CS, Models of HCI, Computational Social Science.

Bachelor's of Engineering in Information Technology

August 2007 - June 2011

Rajeev Gandhi Technical University, Bhopal, India

Percentage: 74.03/100.0
Coursework: Operating Systems, Database Management Systems, Data Structures, Analysis and Design of Algorithms, Computer Networks, Web Technologies, Artificial Intelligence, Soft Computing, Information Security.

Certifications

Deep Learning Nanodegree Program

February 1, 2018

Instructors: Prof. Sebastian Thrun (Stanford University), Dr. Ian Goodfellow (Staff Research Scientist, Google Brain), Andrew Trask (Research Scientist, DeepMind)

An extensive program focused on developing practical skills in deep learning. The nanodegree covered topics like: Keras and TensorFlow, convolutional and recurrent neural networks, deep reinforcement learning and GANs. [view certificate].

Malicious Software and its Underground Economy: Two Sides to Every Story

September 24, 2013

Instructor: Prof. Lorenzo Cavallaro, University of London.

Graduate-level course focused on traditional and mobile malicious software, security threats it represents and state-of-the-art analysis and detection techniques. [view certificate].

Startup Engineering

September 23, 2013

Instructors: Prof. Vijay S. Pande, Prof. Balaji S. Srinivasan, Stanford University.

Introductory course on engineering, design and marketing in the context of mobile web development. [view certificate].

Machine Learning

August 13, 2013

Instructor: Prof. Andrew Ng, Stanford University.

Course focused on supervised and unsupervised machine learning algorithms like: regression, classification, support vector machines and clustering. [view certificate].

Web Intelligence and Big Data

June 06, 2013

Instructor: Prof. Gautam Shroff, Indian Institute of Technology

Course focused on building 'web-intelligence' applications using machine-learning and parallel map-reduce programming to analyze 'big data' such as arising from social media or genomics.
[view certificate].

Microsoft certified specialist

February 19, 2013

Programming in HTML5 with JavaScript and CSS3 (MS)

Exam 70-480. [view certificate].

View My Projects

Projects

Some of my software projects.

Forest Detector

sample-image The objective here was to classify regions of forest within a geographic area using deep convolutional neural networks (CNNs). The project involved development of software infrastructure to train different CNN architectures on a dataset of "forest" and "non-forest" images. The trained networks were then used to automatically annotate forest and non-forest regions in a large aerial image of Floyd County, VA. Technologies used: Python, Pandas, Flask, TensorFlow, Keras. More details available in this paper. Source code for this project is available here.

ApartMates

sample-image ApartMates is a web service that enables delegation and sharing of responsibilities and expenses among roommates in shared apartments. Users can create tasks with deadlines and priorities, shared amongst all roommates, and earn reputation points for finishing tasks on time. The system automatically generates grocery lists and grocery prices using the Walmart Open API. It also provides a RESTful interface to access user reputation points. Technologies used: JavaEE, PrimeFaces, JPA, MySQL, JAX-RS. More details available in the report. Source code for this project is available here.

Beat-Click

sample-image This is a system consisting of Android wearable and smartphone applications which captures images using the smartphone camera when the heart rate (HR) of the user changes. The key idea is certain HR patterns are associated with a change in a person's disposition, and thus can be used as signals to capture important events a person experiences, while doing an activity (e.g., hiking) using photographs. The captured photographs along with associated HR values and timestamps can be analyzed post-hoc to discover, visualize and document changes in user's mood through a hike and the important events associated with these changes. Such information can be used to automatically capture and filter important memories for the user, without disrupting their hiking experience. Technologies used: Java, Python, Android SDK, Wear SDK, Plotly. Details about the system and the broader project are available in this paper. Source code is available here.

Memory Box

sample-image The objective here was to build a system that can find instances (frames) within videos in a large video library that match a natural language user query. The system builds on the Automated Image Captioning Using Deep Learning approach by Andrej Karpathy, Fei-Fei Li and extends it to videos. In essence, the system pre-processes all the videos in the library, by sampling frames (i.e. images) from each video, every few seconds and generates natural language captions for each frame using a Convolutional neural network and a Long short-term memory network. These captions are then indexed as documents along with the corresponding video url and timestamp. When a users submits a search query, this index is used to retrieve the caption that best matches the search terms, along with the video and the precise timestamp, within the video associated with the caption. This project was built during the Build to Learn hackathon at Virginia Tech. Technologies used: Python, Flask, PyTorch, Lucene, CNN, LSTM, Bootstrap. Video demo and more details are available on Devpost. The source code can be found here.

The Social Network

sample-image This is one of my undergraduate projects, a social networking web service. The system application is capable of securely loging users in and out. Users can create public profiles, upload profile picture and chat with other users. It also has the “remember me” feature which allows users to access their profiles without logging in. Technologies used: PHP, MySQL, HTML, CSS, JavaScript, Apache, Google Maps API. More details available in this report. Source code for this project is available here.

A* Maze Solver

sample-image This is a NetLogo toy simulation which simulates the A* path finding algorithm. Users can draw obstacles or completes mazes and the agent (turtle) will find the shortest possible path (if one exists) to a destination point (patch). The path finding is visualized in the model, so users can observe, understand and learn how the algorithm works. More details, including how to use the model are available on the official NetLogo user community page.

My Achievements

Awards

Few of my accomplishments.

TCS ‘Service & Commitment’ award.
TCS ‘Star of the learners group’ award.
All India Rank of '946' in CS GATE 2011 amongst 136,027 candidates.
'First Rank' in paper presentation contest @ UIT Bhopal’s Annual technical festival GECSPARK 1.0.

Organized Technocrats Institute of Technology’s Annual technical festival Technofusion’10.
'Second Rank' in Robo Race held at OIST’s National Conference (Auto Innovation).
Ran the 'Asheville' and 'Hokie Half-Marathons' in 2017.

Contact Me See My Research

Research

Below are my research endeavors.

Social Pathways

This project seeks to incorporate social behavior into mathematical models of infectious disease transmission dynamics, with a focus on influenza like illness. My contribution to this project is an approach to model and calibrate agent behavior in large multi-agent simulators using survey data. The objective is to make the behavior of the simulated agent popoulations close to that of real human populations.

Behavior Model Calibration For Epidemic Simulators. Accepted at, International Conference on Autonomous Agents and Multi-Agent Systems, 2018

Human Data Collection and Behavioral Modelling

Built grounded, fine-grained behavioral models which use "fragments" of behavior mined from past literature in the social sciences as well as studies conducted in the field. Fragments are the building blocks used to compose grounded fine-grained human behavior models. This culminated in a set of guidelines and a behavior composition approach. These were used to model a support services organization with an aim to study the factors that drive productivity and absenteeism in such businesses. The detailed guidelines, the behavior composition approach and the support services organization case study have been published in the Winter Simulation Conference, 2016. These efforts also lead to three other case studies which were published in prominent modeling and simulation conferences and workshops.

Meghendra Singh, Mayuri Duggirala, Harshal Hayatnagarkar, Sachin Patel, Vivek Balaraman: Towards fine grained human behaviour simulation models. Winter Simulation Conference 2016: 3452-3463 [source].
Vivek Balaraman, Harshal Hayatnagarkar, Meghendra Singh, Mayuri Duggirala: Towards Better Crisis Management in Support Services Organizations Using Fine Grained Agent Based Simulation. PRIMA 2016: 366-375 [source].
Mayuri Duggirala, Meghendra Singh, Harshal Hayatnagarkar, Sachin Patel, Vivek Balaraman: Understanding impact of stress on workplace outcomes using an agent based simulation. SummerSim 2016: 35 [source].
Meghendra Singh, Mayuri Duggirala, Harshal Hayatnagarkar, Vivek Balaraman: A Multi Agent Based Human Behaviour Modelling Approach to Enterprise Simulation. ModSym+SAAAS@ISEC 2016: 10-15 [source].

Norms and Organizations

Norms are the policies, rules, explicitly stated or unwritten codes of conduct that has been prescribed to or agreed upon by a society or community to govern their behaviour. The process of norm formation, establishment, spread and function are an area of considerable interest to many disciplines from sociology, psychology, economics to computer science. This project aimed to extend the pioneering work of Robert Axelrod, on The Norms Game. I worked on various evolutionary algorithms which modeled the emergence of norms and there relationships to different organization structures. This resulted in three publications at software engineering and simulation science conferences.

Meghendra Singh, Vivek Balaraman: Examining norm establishment and spread in different organizational structures using an extended Axelrod model. SpringSim (ADS) 2015: 42-52 [source].
Vivek Balaraman, Meghendra Singh. Exploring norm establishment in organizations using an extended axelrod model and with two new metanorms. SummerSim 2014: 39 [source].
Sandeep Athavale, Meghendra Singh: Modeling work-ethics spread in software organizations. CHASE 2014: 2-7 [source].

Contact Me

Hello, hola, नमस्ते !

I'm Meghendra.