Jul 2023 – Oct 2024 · 1 yr 3 mos

• Engineering the streaming platform for Disney+ hotstar with server side ad insertion at scale to support 50 million+ concurrent users on live stream with granular and targeted personalised ads.

Jul 2021 – Jul 2023 · 2 yrs

• Co-led engineering design and architecture for services within the Core Platform enabling a consistent design language across various platforms for hotstar and easing development lifecycles. Tech Lead for Layout Service and co-designed various other components built for high performance and scale along with hands-on development for features within the Core Platform.
• The features built in this platform are now live in 18+ countries for Disney+ in MENA and South Africa

Apr 2019 – Jun 2021 · 2 yrs 2 mos

• Built engineering capabilities to deploy offline, online/near realtime services to enable the personalisation engine for hotstar. Worked on the engineering design for the various product features and tuned for performance. Maintained infrastructure required as code, made data store choices for different query patterns. Debugged several critical engineering bugs - many of which led to impactful throughput optimisations at the organisation level. Implemented and designed changes for personalising layouts across different users and enabled the product teams to A/B test varied layouts across different types of cohorts.

Oct 2017 – Mar 2019 · 1 yr 5 mos · Bengaluru Area, India

• Engineering: Worked as a part of a 4-member engineering team and contributed to building Ola Shadow - a scalable simulator service that mimics Ola business. The simulation infrastructure facilitates data scientists and engineers to check the business impact of their new models and algorithms on a plug and play basis.
• Science:
• Worked on building models and formulated a strategy that rationalizes offers that are rolled out to customers so that relevance, redundancy and ROI are accounted for, in order to keep burn under control. In addition, recommended a slab-based optimal pricing for Ola Pass in order to increase regularity of rides and incur as less burn as possible.
• Worked on designing an algorithm to solve the core allocation problem along with spatio-temporal shaping supply of cabs in order to improve bookings, GMV and utilization of the platform.

Feb 2017 – Oct 2017 · 8 mos

• As a part of the Algorithms Team for Search for Adobe Media Optimizer (Adobe Experience Cloud), I have worked on solving non-trivial optimization problems leveraging additional information from Search Engines like device targeting, location targeting, remarketing targets, hour of day and day of week to re-solve the entire optimization problem for SEM with additional constraints at levels of hierarchies that allows users to set bid adjustments in these targets to ensure better performance of their advertising campaigns.
• Worked on, and deployed a two step optimization framework that works in tandem with the existing framework that gives theoretical guarantees on better revenues than the existing framework.
• Proposed and implemented an iterative EM like algorithm leveraging all kinds of targeting information, making suitable practical approximations that ensure an increased revenue for the clients for the same budgets. [Beta]

Jul 2015 – Jan 2017 · 1 yr 6 mos

• As a part of the Algorithms Team for Adobe Media Optimizer under the Digital Marketing Business Unit of Adobe, my responsibilities include working on:
• Formulating, building and deploying reliable algorithms-based ad optimization systems that work at scale and optimize return on online advertising spend that Adobe manages.
• Implement automated models that use statistical and machine learning techniques.

Jan 2015 – May 2015 · 4 mos

• Graduate Teaching Assistant for the Advanced Programming Lab (CS2810) offered for undergraduate students in their fourth semester.
• Mentored students and designed assignments and projects for the course
• Assignments were designed with an academic and interview-centric approach.
• Topics covered: Advanced Graph Algorithms, Dynamic Programming, Heaps, Network Flows.
• The students were introduced to OOPs through Java and were trained in a way to make sure they can pick up any programming language in the future on the fly.
• Introduction to Multi-threading and designing games with UIs and basic AI were the focus of the projects

Jul 2014 – Dec 2014 · 5 mos

• Led a team of teaching assistants for the Programming Lab - CS2110 - offered for undergraduate students in the third semester.
• Mentored students and assigned mentors for every student in order to bridge the divide in the levels of programming experience of various students in the class prior to college.
• In order to bridge the divide, the focus was to make sure every student would at least be able to (1) code proficiently in C with a coherent coding style, (2) debug code using a debugger and make it a habit to use one, (3) be comfortable with basic tools viz. an editor, a debugger, makefiles, etc., (4) be able to think in C to the degree required to implement pseudocode.
• Designed assignments and projects for the course. Imperative programming being the focus of the course.
• Topics covered: Programming in Unix based environment, Pointers, Linked Lists (Stacks, Queues, it's applications), Writing Libraries (including concepts of separate compilation, makefiles), Recursion, File handling, Binary Trees, Binary Search Trees, Backtracking, Divide and Conquer.
• Student were encouraged to pursue a project, and were mentored for the same, apart from the course assignments.

May 2013 – Jul 2013 · 2 mos · Noida, India

• As a research intern with the ATL (Adobe Research) team I worked on developing new algorithms to measure similarity in PDF documents. Specifically, we developed an offline algorithm to measure image similarity, and another algorithm to measure layout similarity between the documents to be used for other machine learning tasks such as clustering similar documents. We have successfully filed for a patent for our layout similarity algorithm with the US Patent and Trademark Office.

May 2012 – Jul 2012 · 2 mos

• Worked on a Web Socket API, ensured load maintenance when client is inactive and implemented a real time Queue to monitor database activity.
• Implemented a similarity engine similar to the “More Like This” feature using Information Retrieval techniques for the back-end which returns records similar to a document in decreasing order of rank