Avinash Kadimisetty


I am a graduate student in computer science at the University of Illinois Urbana-Champaign, where I work in the field of Machine Learning and Deep Learning.

I interned at Yahoo in the Data team in summer 2019. Previously, I worked as a Junior Data Scientist from January 2017 to July 2018 at Evive Health, LLC. I worked on applying Machine Learning to health data.

I obtained my Bachelor's Degree in Computer Engineering from IIITDM Kancheepuram in 2016.

Email  /  Resume  /  Github  /  Google Scholar  /  LinkedIn



Work Experience

Summer Intern at Yahoo - May to August, 2019

I worked in the data platforms team at Yahoo Champaign office.

  • Designed and developed a product for generating insights from advertising data across all data sources at Yahoo
  • Reduced the workload of the sales team for a customer meeting by 70%
  • Tools/Technologies used: Apache Pig, Hive, Presto, Python, Looker
Achievements: Rank 2 in a Yahoo homepage article recommendation machine learning contest organized across all Yahoo offices in the world.

Junior Data Scientist at Evive Health, LLC - January, 2017 to July, 2018

As a Junior Data Scientist at Evive, I worked on several Data Science projects in the healthcare domain.

  • User Behaviour Modelling: Improved the conversion rates by 13% using ML to target potential users.

  • Medical Event Modelling: Trained predictive models on AWS EC2 instances to find the chance of occurrence of chronic diseases which avoided $2.1mn in unnecessary treatments.

  • Hospital Readmission: Identified users at high risk of readmission using machine learning techniques like CNN, Random Forest, and Logistic Regression to notify them early and avoid huge healthcare costs.

  • Report Automation: Accelerated the report generation process at Evive to reduce the number of analysis hours by 60%.

Trainee Decision Scientist at Mu Sigma Inc. - August, 2016 to January, 2017

As a Trainee Decision Scientist at Mu Sigma, I was involved in the development of the following analytic products.

  • muScrum: Developed a web-app to store and analyze scrum details to reduce bi-weekly sprint analysis hours by 15%.

  • muMix: Added new visualizations to show optimum spends across multiple channels in a marketing mix product.
Projects

  • Neural Image Caption Generator: Implemented a Neural Image Caption (NIC) generator based on the paper Show and Tell in Python using PyTorch framework. Benchmarked its performance on MSCOCO and Flickr datasets.

  • Image Super Resolution: Created an image super resolution framework for x-ray images using Single Image Super Resolution Residual Neural Network. Stood in the top 10 percentile of the class with an average RMSE of 1.41.

  • Relation Extraction from Genomic Text: Developed an engine to construct a taxonomy from genomic text by mining relational patterns from the corpus and then imposing a semantically typed structure on the patterns.

  • Online Recruitment Portals: Developed web portals for Faculty and Staff Recruitment at IIITDM Kancheepuram. Automated selection process reduced the manual collection and selection process by 30%.

  • Student Portal - 4pi: An interactive online platform to share information breaking down all the barriers of accessibility and communication that existed earlier. Main features include Posts, Polls, Events and Professional Profiles, which can be viewed from anywhere across the globe. Designed the entire front end of the portal.
Research

I'm interested in research related to Machine Learning, Computer Vision and Deep Learning. My publications are listed below.

Lossy image compression - A frequent sequence mining perspective employing efficient clustering
Avinash Kadimisetty, C. Oswald, B. Sivaselvan, K. Alekhya
INDICON, 2016

This work explores the scope of Frequent Sequence Mining in the domain of Lossy Image Compression. The proposed work is based on the idea of clustering pixels and using the cluster identifiers in the compression. This method focuses mainly on applying k-means clustering in parallel to all blocks of each component of the image to reduce the compression time.

Frequent Pattern Mining Approach to Image Compression
Avinash Kadimisetty, C. Oswald, B. Sivaselvan
ADCOM, 2016

The paper focuses on Image Compression, explaining efficient approaches based on Frequent Pattern Mining(FPM). Redundant data in the image is effectively handled by replacing the DCT phase of conventional JPEG through a mixture of k-means Clustering and Closed Frequent Sequence Mining.

Teaching

I worked as a Teaching Assistant for the following course.

Cloned from here!