Mohit Surana Developer | ML Enthusiast

About Me

Hi there! I am a Computer Science graduate student at the University of Massachusetts - Amherst.

Intruiged by all things Machine Learning, I love staying up to date with the happenings in the research community. I am eager to transform my theoretical knowledge into practical and scalable solutions through a meaningful career in the industry starting in the summer of 2019. Here is my resume.

My interests lie in the fields of software engineering, large scale systems, and machine learning. There are petabytes of raw data waiting to be processed to obtain invaluable insight and I wish to dive into that pool of information.

My goal is to develop a mechanism by which we can skill the common and underprivileged masses who do not have access to good quality education by using technology to aid or automate the teaching process.

My Expertise


Numerous years of experience through exposure via courses, implemented projects and technical internships


Used Numpy, Scikit-learn and MATLAB to implement core machine learning, natural language processing and vision algorithms

Of late, I have been dabbling in the PyTorch and Keras frameworks, while handling larger workloads through PySpark


Developed simple applications, and dashboards with data visualization to effectively communicate results

Strong proponent of version control tools like Git


I have worked for the following organizations in the past


Designed a deep learning architecture to predict if a timecard will require edits. Focusing on high risk timecards, while auto approving low risk ones can lead to substantial time savings for all customers

Conducted exploratory analyses on support tickets to identify business use cases, configured Solr to index and browse tickets, and used machine learning models to route tickets enabling faster resolution

Stack: Python, Keras, PySpark, Scala, Apache Solr, SQL


Remodelled resources and wrote fully automated end to end tested APIs

Created dashboards for monitoring API performance using Splunk and productivity tracking for Github activity

Stack: Java, Splunk, Django (Python)

Goldman Sachs

Designed a portal to onboard vendors onto an invoice management system

Data federation helped automate of invoice management leading to an estimated reduction of 120,000 customer tickets annually

Stack: Angular Js, RESTful Services, Java and proprietary DB tools

Student Nokia Developer

Backend developer for a multiplayer Movie Quote Quiz game on Windows Phone

Optimised it to be playable on unreliable low speed networks

Stack: Windows Phone SDK

Lab Projects

Information Extraction & Synthesis Lab - UMass

Advised by Dr. Andrew McCallum, Nick, Ari and Rajarshi, I am researching ways to navigate large scale graphs to find nearest neighbours efficiently using Reinforcement Learning for my master's project.

IESL at UMass develops techniques to mine actionable knowledge from unstructured text.

Information Sciences Institute - USC

As part of an industry collaboration project, I worked [github] on the Enron Email Dataset to infer organizational hierarchy and model user behaviour to detect possible phishing attacks in an unsupervised manner

The ISI at University of Southern California is a world leader in research and development of advanced information processing, computer and communications technologies

Center for Knowledge Communication - UMass

Under the guidance of Dr. Beverly Woolf, I worked on the full stack development of a crowd sourced question authoring tool for MathSpring, a Mathematics tutor

The Center for Knowledge Communication at UMass builds Web-based intelligent tutors that understand a student's learning needs, optimize teaching materials and use effective tutoring strategies

Past Projects

Bloom's Taxonomy

Applying Bloom’s Taxonomy in Qualitative Analysis of Questions

Developed classifiers for labelling textbook questions along the cognitive (90% accuracy) and knowledge (60% accuracy) dimensions of Bloom’s Taxonomy.

Design included an ensemble of numerous classifiers

Poster was presented at the UMass Data Science Mixer and the PESIT Roadshow

GitHub DS Mixer Poster Roadshow Poster

Sentiment Analysis of Tweets

Compared the accuracies of a Naive Bayes Classifier and Decision Tree with respect to the Indian Elections

Implemented a Bidirectional RNN based solution to assess the performance of banks after the Demonetization scheme


Intelligent IDE and Python Code Autocorrector

Developed an IDE with features like language based highlighting, indentation and auto completion.

Also wrote the grammar and a Python based parser that checked for syntactic and semantic errors in code, suggesting fixes

Shopping Portal

Online Shopping Portal

Led a team of 13 for a Software Engineering Project

Concretely understood the development lifecycle and effective management across multiple developers

Designed the backend analytics pipeline for suggesting how much should be restocked and when