Newark, California, United States
For over 12 years, I've been at the forefront of engineering innovation at Google (and Oracle), leading teams to design and deliver software products that generate significant business impact. My passion is building and scaling high-performing engineering organizations focused on cutting-edge technologies. I excel at crafting and executing multi-year strategic visions, leveraging my deep expertise in areas like LLM Agents, distributed systems, ML infrastructure, and data reliability to drive tangible results. Impact Highlights: * Delivered XXX+ SWE-years in efficiency gains by leading the development of a YTAds-wide LLM-based agentic platform for automated Ads debugging, faster experimentation. * Engineered multi-tiered automation to protect mission-critical YTAds binaries, ensuring the reliability and security of Ads targeting and auction systems. * Spearheaded a YTAds-wide initiative across 100+ teams, unlocking $XXXM revenue and saving XX+ SWE years annually by accelerating product feature delivery. * Implemented data anomaly detection for 100K+ ML features, enhancing data reliability across Google Ads ML models. * Pioneered a Virtual Google Home environment, eliminating $XM in TVC costs and enabling faster product iteration.
Leading a high-impact team building a central "Nerve Center" for Google’s transition to Intent-Driven Software Development. I lead the overall strategy and development of a unified control plane that shifts the engineer's role from manual execution to strategic oversight (Human-on-the-Loop), effectively bridging the gap between Inner and Outer loop development. Our core architecture includes Context Engines, Intent Management, Skill Management, Dynamic Canvas frameworks, as well as Personalization and Agent Trajectory/Observability—enabling thousands of teams within Google to delegate complex, long-running SDLC tasks to a federated ecosystem of specialized AI agents. By fostering a "Bring Your Own Agent" (BYOA) model, we empower teams to integrate specialized agents tailored to their specific business nuances. In partnership with Google DeepMind, we are integrating these agentic solutions across IDEs, CLIs, and first-party web surfaces to define the 2026 gold standard for AI-native developer productivity and infrastructure at scale.
I lead the Engineering Velocity across YouTube Ads (Team of 15) to achieve faster time-to-market for product features that saves expensive engineering resources (50+ SWE / year) and unlocks $XXM revenue for YTAds. Stack involved includes Ad Retrieval & targeting, Brand Safety, ML Quality, Advertiser Experience, Formats &Data Infra. Key achievements include YTAds wide Agentic workflows using LLM based Agents to achieve 10x reduction in on-call load, 50%+ improvement in experiment launches, 2x faster improvement in addressing advertizer escalations, and automated large scale test generation. Past efforts include YTAds Prevention efforts reducing the overall incidents by order of 10x.
As a hands-on leader and manager, I've led a high-performance ML Infrastructure Automation, Efficiency, and Reliability team of 7, achieving significant results. I launched Training Serving Skew for over 100,000 features and Canary Analysis for online model deployment for 20K models, collectively addressing $90M+ in potential revenue impact. Additionally, I automated unused model/feature cleanup, impacting 50+ SWEs, and implemented feature lineage for DMA support. I've launched 7+ cross-functional products supporting a team of 200+ ML developers across Ads and developed a multi-year strategy for feature monitoring and data reliability for the Ads team.
Establishing Audience one infra as canonical x-Ads user-modeling platform for O&O. Full-stack development using Angular and Scaffolding server for User Model browsing, dependency visualization, resource estimation, Training service and backfill monitoring. Cross functional engagement with multiple teams.
Continuing my work at Oracle as Lead Developer of next-generation in-house high-performance scalable simulation tool.
Developing and Supporting High-performance CAD tools. Adding more features and making it adaptable to the future technologies. Experience include - Formulation, Development, Testing and Deployment of patented Algorithms for faster simulation of huge netlists. - Generation of Accurate and Compact models with very high performance. (3X better than the competitor's tool) - Improved the performance of the legacy in-house tool by * Analyzing and Optimization of C++/C Code using Symbolic Tool Box. (~40% improvement) * Profiling and Parallelizing using OpenMP for Performance (~62% improvement)
Experience Include - * Designed, Developed and Deployed "Get Set Go" in IOS platform. * Working with iOS 8.1 SDK, Objective-C, Cocoa Touch, UIKit, Xcode, Interface Builder, Instruments, and Git on a daily basis * Rapid Prototyping for quick reviews and feedbacks * MVC based Designs * Distributing for testing and launching in the App Store.
Keen interest in developing websites with state-of-the-art technology. Experience includes * Rapid prototyping. * Developing entire website using MVC based Web Framework. * Deploying and Hosting site over Amazon EC2's * Designing fluid pages using HTML and CSS3