Building custom AI agents to automate and enhance enterprise workflows. Passionate about driving innovation and efficiency. Learn more at zangoh.com

I founded Newzera with the vision of creating the world’s most immersive news consumption experience. Newzera is the first short-video news platform where content is produced using proprietary AI. Our human-centered AI scans the web, condensing news into fast, verified, and engaging snackable videos. Designed to be the “Netflix of short news,” Newzera offers a personalized stream of captivating content across a variety of interest categories, making news consumption both informative and entertaining.Through NewzGPT, our advanced AI, we set new standards in news generation, fact verification, and personalization. Newzera became a go-to destination for users seeking a fun, gamified way to stay informed, blending the best of entertainment and news.Key Achievements:- Scaled the product to 20k+ Monthly Active Users (MAUs).- Pioneered NewzGPT, the world’s most advanced AI for news content creation and verification.- Awarded AngelList’s Best Startup in the World and selected for AngelList's Early Stage Quant Fund.- Part of NASSCOM DeepTech Club, supporting deep technology startups.- DIPP certified under Startup India.

At Pinterest, I played a key role in enhancing the pins recommendation system, which drives over 40% of the platform's total user engagement. With 175M weekly active users and handling 65k QPS, this system is critical to Pinterest's success.Key Contributions:Recommendation System: - Led projects to improve candidate flow, ranking, and quality, resulting in a 7.1% increase in clickthroughs and a 5% rise in repins. - Developed and launched three new text-based candidate sources, utilizing pin annotations, search logs, and hybrid search mechanisms.Cold-Start Problem Solutions: - Designed and implemented two new candidate sources based on visual similarity and random walk candidates. - Introduced a new ranking model with pin category features, boosting fresh pins clicks by 42.37% and repins by 21.5%.Revenue Uplift: - Led a project to optimize candidates for promoted pins, leading to a 2.5% sitewide revenue gain, adding $17.5M.Next-Gen Infrastructure Development: - Led the team to develop the next generation of related pins, designing and building the entire infrastructure from scratch, including backend stack, API stack, and monitoring tools.

After Amazon acquired Post Intelligence, I joined AWS to work on the foundations of what was then a top-secret project—now known as AWS Honeycode. My work involved laying the groundwork for this no-code application builder, which enables users to create custom business applications without programming.

At Post Intelligence, I played a pivotal role in designing and developing innovative systems that significantly enhanced the performance and capabilities of our platforms. My work contributed to the company’s success and eventual acquisition by AWS.Key Contributions:Computational Advertising System (US Patent): - Designed and launched a novel gallery-based social ads system using a Bayesian Inference model, increasing RPM by 15% for mobile and 10% for desktop traffic on the MyLikes platform. This work is recognized with a US patent for determining ad ranking and placement [US15/427550].Onboarding Workflow Optimization: - Developed a new onboarding workflow for MyLikes' publisher interface, resulting in a 10% increase in signups.Content Management System Enhancement: - Launched an image upload and thumbnailing feature, improving interface speed by 30% and enhancing bad image detection.Backend Infrastructure Development: - Designed and developed the backend infrastructure for Post Intelligence, focusing on a robust crawling system for content across social media platforms.Deep Learning Model for Social Media: - Created a deep network to evaluate the probability of a tweet going viral, improving content engagement strategies.Content Suggestions Engine: - Developed and launched a content suggestions engine that personalized recommendations based on users' social profiles.

At Epic, I was part of the Orthopedics group, responsible for developing modules related to orthopedics within the EpicCare Ambulatory software. My role involved both technical development and user-focused enhancements.Key Contributions:- Successfully completed a 6-month intensive training program, becoming certified in M programming language, VB6, ActiveX, and Hyperspace.- Contributed to the development of critical modules for the Orthopedics group within EpicCare Ambulatory, enhancing patient care through software innovation.- Designed and developed widgets that provided high-level summaries of patient information, enabling end-users to access vital data and functionalities during patient encounters.

During my time at IIT Kharagpur, I served as a Teaching Assistant, contributing to the academic development of students in key computer science courses.Key Contributions:- Assisted in teaching the course 'Theory of Computation' during the fall semester, helping students grasp complex theoretical concepts.- Supported the course 'Computer Networks' in the spring semester, facilitating students' understanding of network principles and applications.

During my internship at Yahoo, I focused on developing a news recommendation system that leveraged Twitter data to enhance user engagement with Yahoo! News.Key Contributions: - Designed and developed a news recommendation system utilizing user tweets, followers' tweets, trending topics on Twitter, and Yahoo! News. - Created a new algorithm to rank macro categories for an entity using Wikipedia, Freebase, WordNet, and YAGO2 ontology. - Awarded the Best Intern Award out of 20 interns for outstanding performance.

As one of six students selected from across India, I had the opportunity to intern at one of Europe's top universities, École Polytechnique Fédérale de Lausanne (EPFL).Key Contributions: - Developed tools enabling developers to create privacy-preserving applications on the Qt framework. - Created a runtime simulator to generate a ‘privacy score’ for applications, enhancing user privacy protection. - Integrated new features into the Qt IDE, including code auto-completion and tool-tips to rank the least privacy-intrusive functions higher.

During my internship at the Center for Development of Advanced Computing (CDAC), I focused on enhancing the efficiency of process scheduling on a High-Performance Computing (HPC) cluster.Key Contributions: - Modified Hadoop’s scheduling algorithm to utilize idle time pockets on processors for auxiliary processes, improving overall efficiency. - Developed an algorithm to accurately estimate idle times and schedule auxiliary jobs without aborting or purging them. - Increased the efficiency of the HPC cluster with 4,608 cores to 98%.