Tanay Deshmukh

Tanay Deshmukh

Software Developer

Contact Me

About Me

I'm a developer who's passionate about coding and building great products. I graduated with a Master's degree in Computer Science at the University of Southern California. I'm currently working for Amazon in Bay Area, CA. I like to code mainly in Python and Java. My interests lie in Machine Learning, Data Science and Web Development. Check out this Chrome extension I made with a couple of friends to filter and flag extremist webpages at AngelHack 2017 in Dubai. We won the first prize and $500 at the three day event.

I love to play video games in my free time; I'm a huge fan of the Dark Souls series. I like to play the guitar and read too. I have an unhealthy addiction to Reddit and Coke Zero. Hit me up if you're looking for developers to build something exciting and let's get coffee together!

Latest Projects

Pronto Auto Login

A Chrome Extension to automatically login to VIT University's Wifi Captive portal

Pronto Auto Login

Developed a Chrome Extension to automatically login to University Wifi network which had a captive portal. The extension has hundreds of weekly users and a 4.7 star rating.

Find out more

Flag a Spot

Flag a Spot

A crowdsourced crime-reporting Android application with a Parse+node.js backend. Uses an MVVP-based architecture with EventBus making it highly decoupled and scalable. Utilizes crowdsourced reports of criminal activity to build a map of crimes in an area and sends reports to users if area is unsafe.

Find out more

FExtrimism

FExtremism

A Google Chrome extension with a node.js backend for the efficient filtration and flagging of webpages linked to on social media containing violent extremist propaganda, identified through composite parameters. Won the GovTech Alliance prize at AngelHack 2017 Dubai.

Find out more

Travel and Entertainment Search

Travel and Entertainment Search

Developed a responsive web-application using AngularJS + Bootstrap on frontend and node.js+NGINX on AWS as backend. It uses data from Google Places and Yelp APIs to return nearby places of interest to the user based on current location.

Work Experience

Software Development Engineer II Amazon (December 2021 - Present)

  • Currently working on AI related tooling at Amazon.
  • Worked on the Widget Gallery feature on the new Echo Show 15 device, and later on other Echo Show devices.
  • Served as Scrum Master for the team for over a year, leading the sprint grooming and planning meetings as well as running the daily standups and weekly retrospectives.

Software Development Engineer I Amazon (July 2019 - December 2021)

  • Developed features to drive customer enagement in the Alexa personality domain.
  • Working on backend systems to write robust scalable code in Java and Kotlin.

Student Research Assistant - USC Information Sciences Institute (June 2018 - May 2019)

  • Worked on the DSBox project which implements the D3M API developed by DARPA.
  • Implemented a Date Featurization module for the project which could identify and extract dates from a dataset with 99% accuracy.
  • Developed an algorithm for record linkage using the Record Linkage ToolKit developed by ISI. Obtained an F-score of 55% which was better than current state-of-art methods.

Data Science Intern - Kristal.ai (January 2017 – May 2017)

  • Worked on building a recommendation engine to match investors with investment portfolios (Kristals). Used an SVM model and implemented in TensorFlow + scikit-learn.
  • Devised and implemented an algorithm to determine the similarity of any two portfolios and accordingly rank them.
  • Wrote web-scrapers in Python to scrape financial data from websites like Bloomberg and Interactive Brokers. Scraped over 30GB of data which saved the company a lot of manual labour.
  • Vastly improved on existing chatbot system, which was made by an external contractor, by building an in-house chatbot using IBM Watson API in less than a month.

Software Development Intern - RedGirraffe (June 2016 – July 2016)

  • Developed a Recommendation Engine for real estate properties in Python using scikit-learn.
  • Deployed a RESTful web service for the same using Flask+NGINX.
  • Used a collaborative filtering strategy by employing k-means clustering. Ideal due to lack of training data.

Other Projects and Publications

Context-aware Clustering using GloVe and k-means

Developed a novel context aware clustering algorithm based on k-means using GloVe and Word2Vec word embeddings. Published in International Journal of Software Engineering & Applications (IJSEA), Vol.8, No.4, July 2017

Link to paper

Continuous Detection and Screening of Epileptic Seizures

Developed a device for continuous screening and detection of epileptic seizures using Arduino and EEG sensors. Built ML models on Tensorflow and SVMLight. Achieved an accuracy score of 99% and a recall score of 91%. Paper on the same published in Springer conference.

View on GitHub