At the precipice of graduating with a Master's in Computer Science from the University of Massachusetts at Amherst, I am in the market looking to employ my skills in Machine Learning and Software Development. Consequently, I am seeking full-time positions starting in December 2020/January 2021.
In my SOP for MS admissions, I had written that we are living in the times of the Fourth Industrial Revolution. The world is being accelerated into an age of unimaginable growth that is being enabled and driven by the ease of running sophisticated algorithms on machines of scale, both in terms of numbers and processing power, and by being able to leverage machine learning and AI to truly drive data-driven decision making. While there are many drivers of the revolution, I believe that data science is at the helm of what's accelerating us into the future and I want to be part of the effort that takes us there.
UMass afforded me the opportunity to pursue a Data Science concentration in their Master's Computer Science program. Here, I had the privilege of imbibing knowledge from some of the most revered minds in academia. I worked on projects ranging from full-stack development to creating ensemble methods and ML pipelines. I have learned so much yet there is so much to learn. But I now have the skills and the know-how to deliver value and that is what I am currently seeking- a place for me to employ my skills to provide value.
-
Experience
Stanley Black and Decker - Graduate Machine Learning Researcher Jan 2020 - May 2020
Worked on building an ML pipeline consisting of an unsupervised neural attention based aspect extraction model and VADER sentiment analysis model to extract product features and their sentiment on reviews in the SBD portfolio.
Presented findings via a spider chart and textual qualitative views backed by visualization studies to Data Scientists at SBD.
The pipeline is used in guiding product development/iteration decisions and direct attention towards relevant features.
UMass Amherst Center for Data Science - Full Stack Developer Oct 2019 - Feb 2020
Developed a framework with Bootstrap, Flask, and Redis as a message broker to host an object detection model online.
Project was part of the DS4CG program at UMass for ‘The Nature Conservancy’ to detect the presence of animals in images.
Indian Institute of Management, Lucknow, India - Intern June 2017 - July 2017
The internship was designed to get exposure in R and apply regression models to real-world data.
Developed linear regression models in R, one using Boruta feature selection, to analyze and solve 3 Harvard Business Review case studies (Dean’s Dilemma, MBA Starting Salaries, Managing Employee Retention) with satisfactory results in all.
Zenoti, Hyderabad, India - Intern (Products) May 2016 - July 2016
Converted legacy BimeSync scripts from C# to Python.
Designed and built the architecture for automated testing of OLTP servers and Redshift, after code changes/commits, to ensure backward compatibility. The tool was designed to capture inconsistencies in data and group it for easy correction/refactoring.
The tool was deployed into production and helped reduce man-hours spent on the task.
-
Projects
MapReduce Implementation from Scratch in Java Sept 2020 - Oct 2020
Implemented a distributed system on a single server mimicking MapReduce with 1 master process and 2N worker processes.
Employed reflection to obtain UDFs, hosted sockets as threads to communicate with master, and tolerate faults in workers.
Playing Card Detection in Real Time Sept 2019 - Dec 2019
Performed transfer learning to detect playing cards under ~75ms and over 99% accuracy by training YOLOv3 on custom data.
Generated own labelled dataset of 50000 images of playing cards from 52 images using OpenCV and image augmentations.
Natural Language Text to SQL Sept 2019 - Dec 2019
Analyzed performance of transfer learning on the SPIDER dataset using text2SQL model SQLNet as the baseline.
Built an API on Flask to showcase the model on a website with some results being boosted by up to 10%.
Abstractive Text Summarization Jan 2019 - May 2019
Implemented an attentional sequence to sequence model in conjunction with a pointer generator network and a reinforced actor-critic reward policy gradient algorithm to generate abstract summaries on CNN/Daily Mail dataset.
Received a ROUGE-1 score of 13.3.
Dataset Size Reduction Jan 2018 - June 2018
Replicated a python application based on the paper 'Coresets for Scalable Bayesian Logistic Regression' to construct coresets of different real world and synthetic datasets.
Programmed a jupyter notebook module to implement Principal Component Analysis to reduce feature size.
Big Data Based Security Analytics for Protecting Virtualized Infrastructures in Cloud Computing Apr 2017
Studied and presented an IEEE paper in a technical seminar at MIT, Manipal. Received mention of best speaker.