Feb 2020 – Mar 2024 · 4 yrs 1 mo · Boston, Massachusetts, United States

• Optimized rendering performance of floor maps (warehouses and robot movements) by 25% through Canvas rendering optimizations. Significantly improved user experience by eliminating unnecessary asset re-rendering.
• Created a data pipeline solution that leveraged data from floor entities. This data empowered a machine learning model to predict floor efficiency drops and on-floor issues, providing users with actionable suggestions through the Android app. Resulted in a 7% increase in floor productivity and a reduction in daily delivery times for over 600 million packages.
• Lead the redevelopment of a critical legacy backend service responsible for aggregating, processing, and storing robotic floor data. This project will play a pivotal role in rendering and controlling floor states.
• Design and develop features for Android applications, websites, and backend services utilized by over a million Amazon warehouse employees to visualize and manage robotic operations.
• Rebuilt legacy Android app using MVVM architecture, incorporating advanced libraries like Volley and Guide to boost its capabilities.

Jun 2019 – Aug 2019 · 2 mos · Cambridge, Massachusetts

• Hadooker (a portmanteau for Hadoop + Docker): Developed this feature for creating Hadoop clusters using Docker images and integrated it in the Vertica’s stress test framework. This feature eliminated the need to reserve multiple servers to just one.
• Dynamic Data Generation Tool: Designed and developed this tool to generate parquet/orc files with configurable knobs for tools (Hive, Impala, Spark) to be used, amount of data etc. This helped the QA team unearth various data specific issues while scaling very easily.

Jan 2019 – Jun 2019 · 5 mos · Greater Boston Area

• Created a ”Write a Review” feature with social reviews on Android, increasing user engagement and content creation for 10M+ users.
• Implemented network disruption simulation via Chaos testing on Android, boosting product resilience, reducing downtime, and enhancing user trust.
• Enhanced user experience with seamless horizontal and vertical scrolling in a tile-based layout, optimizing content accessibility.

Aug 2018 – Dec 2019 · 1 yr 4 mos

• Assisted Prof. Leon Reznik on an NSF Funded project (ACl-1547301) for "Data quality and security evaluation framework for mobile devices platform" using machine learning.
• Worked on improving the project for detecting colluded Android applications using RNN's in TensorFlow lite. (Research paper for this project will be published in 2020 IEEE WCCI)

Aug 2018 – Dec 2019 · 1 yr 4 mos

• Teaching Assistant for "CSCI-603 Computational Problem Solving (Data Structures)"
• Grading Students Lab and Assignments
• Presenting Tutorials every week
• Holding Office hours

Aug 2018 – Jan 2019 · 5 mos

• Troubleshooting, performing root cause analysis, and resolving production issues from the application and network layers all the way down to the system level. This might include anything from digging into source code (our own or from open source projects), hunting memory leaks, tracing bottlenecks in upstream networks, or database query optimization.
• Develop new features for frontend and backend.

Jun 2018 – Sep 2018 · 3 mos · Rochester, New York Area

• Created iCitizen’s mobile applications from scratch and development of new features for them.
• Created microservices to fetch data from the users and API endpoints.

Jan 2014 – Jul 2017 · 3 yrs 6 mos · India

• Responsible for design, develop, modify, debug and maintain high volume, mission-critical software solutions in the Telecommunications domain.
• Worked on projects for North American Telecom giants such as T-Mobile(USA) and Rogers(Canada).

Aug 2012 – Aug 2013 · 1 yr · India

• Rainfall Prediction using wind and air moisture parameters collected in real time using GPS and Xbee technology (Digital Radiosonde): In simple words, this project involved designing and making a digital radiosonde using GPS, XBee transreceivers and Arduino board along with collecting the NMEA GPS data and using it to track the course of the balloon attached to it, thus depriving the wind parameters and air humidity and moisture parameters using various capacitive sensors. By feeding this data along with the institutes archived data to the ID3 (Iterative Dichotomiser 3) algorithm we could predict rainfall with 72+% accuracy.
• System for pollution monitoring and assessment: Made a Data acquisition and plotting system using a plethora of sensors, raspberry pi and AWS Iot for pollution assessment and broadcast warnings in case the pollution levels increase a maximum threshold.

Nov 2011 – Jan 2012 · 2 mos · Khadki, Maharashtra, India

• * Worked on a electronic equipment which was used in Infantry Combat Vehicles (ICV).