Oct 2024 – Present · 1 yr 8 mos

Jul 2022 – Dec 2024 · 2 yrs 5 mos

• Led a team of 10+ software and ML engineers in Microsoft Bing, driving innovation and excellence in product development.
• Managed three segments – App Answer, Tech Help, and Events Answer – overseeing end-to-end processes from feature ideation to production deployment. Created and maintained comprehensive engineering system designs from scratch for all three segments.
• Achieved a 4x increase in daily active users (DAU) for App Answer, growing from 500k to 2M in a single year, while maintaining top 3 quality answers in Bing.
• The three segments have a total shown DAU of 10M and consumed DAU of 4M.
• Led multiple partnerships and features, resulting in $5M revenue and 1M new to Bing DAU.
• Received a Manager Signal Score of 93 in April 2024, reflecting exceptional leadership and team management skills.
• Led multiple data science projects in NLP, deep learning, and LLM, including classification, data enrichment, keyphrase extraction, query-document matching, content generation, and ranking.
• Authored papers selected for the Microsoft MLADS (Machine Learning And Data Science) Conference, contributing to the field of machine learning and data science.
• Skilled in strategic planning and execution, driving projects to successful completion within deadlines and budget constraints.
• Experienced in hiring, mentoring and developing team members, promoting continuous learning and professional growth.
• Adept at stakeholder management, effectively communicating project progress and aligning goals with organizational objectives.

Mar 2022 – Jul 2022 · 4 mos

Sep 2018 – Mar 2022 · 3 yrs 6 mos

•  Created multilingual events webpage classifier and automated entity extractor for top 40 Billion Bing webpages
•  Developed and Shipped keyword and category-based triggering approach into production which led to 80% coverage increase in Bing Events Queries triggering. The model had 95% precision and 97% recall.
•  Created multilingual events city scenario classifier and category classifier for all Bing queries.
•  Created vector match based solution for eventname triggering.
•  Created conflation models for events, venues, recurring events.
•  Used BERT and neural network classifiers to predict 35 groups with >95% recall and >99% precision.
•  Received Awards For Excellence in Innovation in Q4 of FY 2020, FY2021 and Spot Award.
•  Mentored summer interns in FY 20 and FY 21 in data science projects.

Jun 2017 – Aug 2018 · 1 yr 2 mos

• 1) Developed and Shipped keyword and category-based triggering approach into production which led to 80% coverage increase in Bing Events Queries triggering.
• 2) Used BERT and neural network classifiers to predict 35 groups with >95% recall and >99% precision.
• 3) Received Awards For Excellence in Innovation and Spot Award

Nov 2016 – May 2017 · 6 mos · Gurgaon, India

• Worked on multiple projects applying the concepts of machine learning, neural networks, computer vision, NLP.
• Tag Recommendation :- Designed an automatic unsupervised keyword extraction algorithm for news articles tags for English and Indian languages.
• Image Based Apparel Recommendation :- Designed an algorithm which upon inputting an image of an apparel, outputs similar apparel images.

Jul 2015 – Jul 2016 · 1 yr · Bengaluru, Karnataka, India

• Worked in the field of predictive modelling,business analytics and operations.
• Started preferential allotment to drivers based on cancellation probability reducing driver cancellation by 40%.Used logistic regression for the same.
• Modeled Driver attrition to successfully predict which drivers would be leaving the system within next 7 days. Implemented decision tree algorithm to get AUROC of 0.9
• Created dashboards by using Hive and SQL database on MSTR .
• Launched Ola Auto in 10 cities.

May 2014 – Jun 2014 · 1 mo · Mumbai Metropolitan Region

• Created Iris Recognition and Fingerprint Recognition systems.
• Iris recognition :- Developed and implemented Iris recognition algorithm in MATLAB. The algorithm took 0.5 seconds to recognize a single iris working with an accuracy of 99.9 %. There were no false positives.
• Fingerprint recognition :- Developed and implemented fingerprint recognition system in MATLAB. The system worked with 90% accuracy and took 0.4 seconds for a single match

May 2013 – Jun 2013 · 1 mo · Bangalore, India

• Made an android application and website capable of scanning QR Code, taking inputs from the user, connecting them with database and then publishing the results on social networking sites.
• Planned the development of product from manufacturing to market and designed it.