Harsh Tyagi

Senior Software Engineer @ Google Deepmind | Making Unified Serving Stack for Gemini Models

Mountain View, California, United States

About

I am a curious engineer focused on scaling the future of AI. Currently, I work on Machine Learning Core Infrastructure, ensuring the global availability and performance of Google’s Large Multimodal Models. I am a key contributor to the unified serving stack for Nano Banana, Gemini’s state-of-the-art image generation and editing model, and Gemini 3/Gemini 3.5 Flash - across Vertex AI, AI Studio, and Gemini. Technical Impact • Google DeepMind: Scaling inference engine for Gemini 3, Nano Banana -and any new model by Google Deepmind in the future. • Systems Infrastructure: Previously built embedded/software systems for Google’s BMC team. • Amazon: Engineered "Buy with Prime" retail widgets, Ads traffic initiatives, and Transportation Optimization engines. • Education: M.S. in Computer Science from the University of Southern California. I am a collaborative teammate driven by the belief that thoughtful software architecture is the key to technological evolution.

Experience

  • Senior Software Engineer at Google DeepMind
    Dec 2025 - Present · 10 mos

  • Software Engineer at Google
    Aug 2022 - Present · 4 yrs 2 mos

  • SDE at Amazon
    Feb 2021 - Aug 2022 · 1 yr 7 mos

    Created software for teams like Transportation Optimization that takes care of optimizing route planning at amazon delivery. Also worked on creating software for "Buy with Prime" widget, which is the new pillar of Amazon retail.

  • SDE Intern at Amazon
    May 2020 - Aug 2020 · 4 mos

  • Student Research Programmer (USC Grad Student) - Internet of Things, AR at Immersive Media Lab- University of Southern California
    Nov 2019 - May 2020 · 7 mos

    Created AR application catering to environmental challenges. Responsible for designing a complete end to end solution for Internet of Things and App Development on mobile and web based platforms Tech Stack: Node.js, C#, Unity, Mapbox, AWS