Kunal Punera Email & Phone Number

Mountain View, Ca, Us

Sunnyvale, Ca, Us

I built and ran 3 AI/Data teams that owned all ML-powered features in LinkedIn Recruiter, Premium, and Learning products; accounting for multi-billion dollars in annual revenue for LinkedIn. The initiatives in my org spanned search and recommendations to some of the first use-cases of Generative AI in the company. One of my extra-org contributions was to help reorganize the engineer evaluation and promotions process across LinkedIn AI.

I started and grew 3 engineering teams with a total of 24 engineers from scratch. My engineering teams were charged with building all the social features of Android Messages. Android Messages is the default messaging app on Android, with hundreds of millions of MAUs. Some of the projects that my team built end to end involved new ways for users to communicate with each other: high quality media sharing, message reactions, message replies, GIF/Stickers/Emoji search, etc. These features involved changes at all levels of the stack, including the RCS protocol.One of the key projects we worked on was helping users express themselves better in communication apps. We used techniques in machine learning and augmented reality to help users create content that they can then share with their contacts. Our team powered all live camera AR effects in the Duo, Android Messages, and Google Keyboard apps.I also initiated and funded a plethora of efforts on code quality, system performance, and release/bug health. As a result of this effort, the engineers are much more productive and Android Messages is much more polished and performant than when I started at Google.

I co-founded Bento Labs and raised the seed round from First Round Capital, Social+Capital, GV, Susa ventures, and some angels. I hired, grew, and ran the engineering team; our team members had worked at Google, Yahoo, RelateIQ, Coursera, VMware etc., and they were passionate about improving the phone experience for billions of people around the world.At Bento Labs we were building an AI aware distribution of Android. Our position was that while many aspects of the mobile OS were commoditized (touch, sensor, battery management etc) there was still significant room for innovation higher up the stack, especially in user attention management. While thephones in 2015 had 50x the CPU power, a lot more sensors, and an order of magnitude more bandwidth available to them than a decade before, there still were no good solutions for app discovery, data sharing between apps, or instant personalization on the phone. We believed that if we rethought the app model — what it means to be an app on the phone — and imbued the OS APIs with AI, we could solve some of these long-standing issues. Our first app was Scout Launcher (home-screen on Android) that figured out what information/services the user needed and when they needed it, and provided them right on the phone's home-screen. Our app was highly rated on the Play store (> 4.5 stars) and was used for > 30 mins everyday by its users.

- helping them with their machine learning algorithms

I was the 4th engineer at RelateIQ and I led the Data Products Engineering team; I built data products that mine and deliver relationship intelligence within the app. Some of the products I worked on are below.1) Automated suggestions to follow-up with contacts based on context of communication (including content of emails). 2) Contact management features such as contact-deduplication and extraction of contact information from email signatures.3) Algorithms to automatic construction of a sales process; such as automatically determining when to add a contact to a sales account. 4) Automatically determining the closest connection of the sales team to a sales lead.5) Machine learning models for parsing emails, extracting content from emails and webpages, and measuring strength of relationships.The techniques I used included various kinds of unsupervised analysis, natural language processing, some supervised learning, and lots and lots of hacks.

Sunnyvale, Ca, Us

- Measuring Impacts on User Engagement: I led this project in which we developed a system to measure the impact of user-experience changes on long-term engagement of users. Our emphasis was on deriving causal links relying mainly on pre-existing (observational) log-data; i.e. without having to run randomized experiments. The first study quantitatively characterized the impact of spam-exposure on long-term user-engagement.- Modeling Social Systems: I extensively researched social system dynamics: modeling the reasons why users link to each other, the propagation of trends, and the production of content on social systems.- Mining Hidden User Intent: I worked extensively on modeling the observed behavior of users to mine their hidden intents. This work resulted in systems to help determine the user intent in ambiguous web search queries (mining facets), in computational advertising systems (commercial intent in queries), and social systems (the reason behind linking).- Web Search Systems: I worked on many layers of the web-search technology stack: from devising low-level indexing and early-termination optimizations all the way up to improving the display of search results (summarizing tweets, shortening page-titles, generating quicklinks and query-suggestions, etc.)- Abuse Detection and Prevention: In this long-running project I developed systems for mail- spam detection, detection and prevention of phishing, design and stress-tests of CAPTCHAs, and data mining of logs to detect abusive behavior.