Principal Applied Scientist at Microsoft | Microsoft 365 Copilot
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Stony Brook University
Master of Science (MS), Computer Science
January 1, 2014 – August 1, 2016
RV College Of Engineering
Bachelor of Engineering (BE), Electronics and Communications Engineering
January 1, 2007 – June 1, 2011
Microsoft
Principal Applied Scientist
June 1, 2026 – Present
Redmond, Washington, United States · Hybrid
Amazon
Senior Applied Scientist
July 1, 2021 – May 1, 2026
Greater Seattle Area
Amazon
Applied Scientist
September 1, 2016 – June 1, 2021
Greater Seattle Area
Natural Language Processing Lab, Stony Brook University
Research Assistant
January 1, 2015 – August 1, 2016
United States
Analyst, Ad Traffic Quality Operations
July 1, 2011 – May 1, 2014
Greater Hyderabad Area
Word vectors for 4th grade processes using Skip-gram
December 1, 2015 – January 1, 2016
Implemented a Torch based system to train word vectors for 4th grade vocabulary. This system trains word vectors using skip-gram (uses word2vec to initialize vectors).
Semantic Role-based Process Knowledge Acquisition.
September 1, 2015 – January 1, 2016
Designed and developed a method to acquire process knowledge using collective role inference across sentences. Semantic and lexico-syntactic features are used for Semantic Role Classifier. Collective role inference is performed using ILP with the help of role classifier and textual similarity scores. The proposed method acquires high quality process knowledge (F1 of 0.72 points). [To be submitted to ACL 2016 for publication.]
The Metis Challenge: Naive Bees Classifier - Driven Data.
September 1, 2015 – Present
Designed and implemented a CNN architecture using Theano to determine the genus of the bee based on photographs of the insects. Achieved an accuracy of 81.9% with just 25 iterations of training on CPU. World Rank: 28
tMood
April 1, 2015 – Present
This project is developed during Bitcamp Hackathon. A Pebble app that analyses the sentiment of people around you using twitter feeds at your location. Backend for this project is developed in Python and hosted in pythonanywhere.com with Frontend developed in JavaScript.
Neera
April 1, 2015 – May 1, 2015
Neera is Rubiks Cube Solver Robot, designed and built using LEGO Mindstorms NXT and programmed using NXC language.
Question Answering System
February 1, 2015 – May 1, 2015
The task for this project involves answering 4th grade process recognition questions by frame alignment. The system takes a 4th grade process recognition question with answer choices as input and produces the correct answer along with the confidence as the output. - Baseline Solution: Finding word overlap (similar to BOW) between question and answer to get word overlap score and ranking the answers based on this score. This answered 48.02% of the questions(152 questions) - Role Based Aligner: First, question and answer sentences are processed using a Semantic Role Labeller to get semantic roles. Then the roles are aligned by measuring the entailment between the entities in the roles to form a alignment score, which is later used to rank answers. This answered 62.50% of the answers. - Jacana Word Aligner: Question and answer sentences are sent directly to Jacana Aligner which employs semi-markov CRF to align the sentences and produce alignment along with a score. This is then used to rank answers. This answered 44.07% of the questions. (The low accuracy of this method is due to the data used in training the CRF. The domain of training data was different than the domain of the questions and answers.)
Predicting facial beauty using Convolutional Neural Network (CNN)
August 1, 2014 – December 1, 2014
In this project, we designed and trained a Convolutional Neural Network (CNN) to predict facial beauty. We configured a deep learning system based on Caffe (a fast framework for deep learning) and trained it to classify faces into one of the 5 classes based on facial beauty. The trained model predicts facial beauty with an accuracy of 50.32%.
Predicting Yelp rating stars from review text
August 1, 2014 – December 1, 2014
The goal of this project was to predict reviews' star ratings on Yelp from review text. We built the following models that perform text analysis on review data to predict the rating stars. 1. Baseline Model: The most common rating, 3 stars, is the rating predicted by this model for all the reviews. 2. Term Frequency Model: In this model we use frequency of word occurrence to predict the review rating. 3. LDA + Sentiment Model: This model predicts rating using Latent Dirichlet Allocation (LDA) with an added sentiment layer by extracting topics and sentiment associated with the review from review text. 4. NMF + Sentiment Model: In this model, we predict review rating using Non-negative Matrix Factorization (NMF) with an added sentiment layer by extracting topics and sentiment associated with the review from review text. We achieved an accuracy of 61% in predicting review rating stars.
Predicting the Super Bowl and College Football Champions of 2015
August 1, 2014 – December 1, 2014
The goal of this project is to build machine learning models to predict the winners of 2015 Super Bowl and the College Football Championship using historical data. We have predicted the outcome of football matches entirely using the knowledge of previous game statistics. We have used three different models to do this: 1. Baseline model : "Point Score Difference Model". In this model we use the score difference to predict winners of future games. 2. Linear Regression Model: In this model, we use linear regression to predict the point difference for each game. 3. PageRank Model: Here, we model the game data as a graph with nodes as teams and edges as score differences between the teams. We then use PageRank on this game graph to rank all the teams. This ranking is used to predict winners of future games.
Development of Controlling and Stabilizing Algorithms for a Quadcopter using LPC2148 ARM Controller
January 1, 2011 – June 1, 2011
The aim of this project was to build a low cost prototype quad-rotor and come up with system stabilization algorithm for the stabilizing the quad rotor. - Came up with a mathematical model of a quad-rotor to correct pitch and roll errors by changing motor thrust. - Used complementary filters to filter noisy sensor data from the inertial measurement unit and get meaningful information. - Was awarded the first place at R.V. College of Engineering for electronics undergraduate research project in May, 2011. - Based on this work, presented a paper entitled “Design of Quadcopter with closed loop control system” at Visvesvaraya Technological University, Belgaum, in 2011.
Project Vyoma
August 1, 2010 – June 1, 2011
Project Vyoma is a research effort at R. V. College of Engineering to design unmanned aerial vehicles. - Devised and built a tachometer based on a rotary encoder which could measure the speed of an IC engine up to 300 RPM. - Designed and developed the DaQ system which captured data during flight from various sensors like GPS, pressure sensor and inertial measurement unit into a SD card to be retrieved later for analysis by aerodynamics team. - Achieved 8th position among 70 teams from all over the world in the 'SAE Aero Design' event held in Atlanta, in 2011.
Cultural Fit Analysis
The candidate demonstrates a strong cultural fit for a data-driven, innovative environment. Their extensive project portfolio showcases a passion for applying advanced analytical and machine learning techniques to solve diverse problems, from bee classification to autonomous agents. The progression through roles at Amazon and Microsoft, coupled with research experience, indicates a drive for continuous learning and contribution to cutting-edge technology. The variety of projects, including hackathons and academic research, suggests a collaborative and curious mindset. The 'Analyst, Ad Traffic Quality Operations' role at Google also aligns with data investigation and quality assurance, which are critical for data analyst roles.
Soft Skills & Operational Fit
The candidate's project descriptions indicate strong problem-solving abilities, a research-oriented mindset, and the capacity to work on complex, multi-faceted problems. Experience as a 'Science lead' and 'Founding member' suggests leadership and initiative. The diverse range of projects, from hackathons to academic research and industry roles, points to adaptability and a proactive approach to learning and development. However, without direct assessment data, specific soft skills like teamwork, communication style, or stress handling cannot be definitively evaluated.