Jul 2022 – Apr 2025 · 2 yrs 9 mos

• Video Ranking on Google Search: short video query intent modelling, unifying long and short video ranking, video deduplication, video segments.

Sep 2021 – Jun 2022 · 9 mos

• Tech Lead in the Ads Privacy Centric Measurement, worked on Safari Private Click Measurement for Google Ads and developed bespoke conversion modelling solutions for novel search/android products.
• This was a relatively fast moving team, so learnt a lot in a short time.
• Had the best manager ever - moved teams to pursue my goal of working on Ranking and ML.

Jul 2018 – Aug 2021 · 3 yrs 1 mo

• Lead Pre-Launch Report and built Loading Time & Abandonment Analytics for Android game developers.
• Also worked on Android Performance Tuner and Peer Benchmarks for Android Vitals.
• Gave a public talk on Vitals @ Google for Mobile, India 2019.

Jun 2016 – Jun 2018 · 2 yrs · Bengaluru

• I worked as a quantitative strategist/data scientist in Divisional Strategies and Global Liquidity Products, within Securities Division.
• My work revolved around two areas: developing models for assessing the liquidity risk for the entire securities division, and optimal deployment of excess cash in liquid assets.

Jan 2016 – May 2016 · 4 mos

• I served as a TA for the Intro. to Machine Learning course, taught by Prof. Sumeet Agarwal.
• I was responsible for designing and evaluating assignments, and conducting one-on-one assignment evaluation & feedback sessions.

Aug 2015 – Jun 2016 · 10 mos

• Multimedia Lab:
• Multi-class Land-cover classification from Satellite-based hyperspectral images (~200 frequency bands).
• I worked on semi-supervised, manifold based classification; 3D texture features; unsupervised image segmentation; and ensemble methods to solve this problem.

Jul 2015 – Dec 2015 · 5 mos

• I was a TA for the Intro. to ML/Pattern Recognition course, taught by Prof. Jayadeva.
• Apart from helping with evaluations, I mentored 4-5 student teams, guiding them through semester-long course projects, some of which were Kaggle competitions.

Aug 2014 – May 2015 · 9 mos

• Data Analytics and Intelligence Lab:
• Topic Modelling from multiple documents/articles about a news event.
• I worked on bespoke Hidden Markov Models with a Dirichlet prior to solve this problem and explored the performance various estimation techniques (variational bayes, gibbs sampling, etc) when given a sparsity-inducing prior distribution.

May 2015 – Jul 2015 · 2 mos · Mountain View, California

• I worked on Diversity Categories for Google News, trained text classifiers using Dist-Belief (the precursor to tensorflow at Google Brain) and created an automated pipeline for extracting & pre-processing data and training such classifiers.
• Diversity categories are topic-agnostic descriptive labels for news articles, such as 'in-depth', 'opinion', 'factual', etc, that inform readers about the different type of articles available on a new event. Built a demo news page with a modified ranking to boost diverse news articles about an event using the above categories.

May 2014 – Jul 2014 · 2 mos · Gurgaon, India

• I worked on object detection and classification in unstructured, crowd sourced images. The problem was to create a good detector-classifier for electricity transmission equipment in crowd-sourced images, with no strict conditions on the size, orientation and number of objects in an image.
• Learnt about and used pre-deep learning image features with a patch based bag of words model.

Aug 2012 – Aug 2013 · 1 yr

• I recruited & led a team of students developing a Rover for a Mars-like environment.
• Helped organize the first astronomy festival in Delhi, the Astro Week, in September 2012 and lead an outreach program to Delhi University, helping achieve a footfall of over 300 students for the festival.