Nadeesha Amarasinghe

Machine Learning at Sunday | ex Tesla AI, NVIDIA, Apple

San Francisco Bay Area

About

Technical leader and core contributor/initiator of many of the ml systems and infrastructure that power Tesla AI (Optimus and FSD). Including large scale distributed training infrastructure, model export and quantization, GPU performance optimization, data infrastructure and research tooling. The models produced (FSD V13/14) have been shipped to millions of vehicles and driven billions of miles. I’m currently working on robot learning at Sunday. Passionate about building ml systems, models and products that positively impact people’s lives.

Experience

  • ML and 🤖 Infra at Sunday
    Jul 2025 - Present · 1 yr 1 mo

    Leading infrastructure and systems software for Machine Learning and Robotics. Join the team https://www.sunday.ai/careers

  • Tesla (7 yrs 2 mos)
    • Engineering Lead, ML Infrastructure
      Dec 2023 - Jun 2025 · 1 yr 7 mos

      Building the machine that builds the models. Lead ml training and inference infrastructure (15+ engineers) for Autopilot and Optimus.

    • Engineering Lead, Inference
      Sep 2022 - Dec 2023 · 1 yr 4 mos

      Started and lead a team (5+ engineers) working on all things AI Inference for Autopilot and Optimus. We worked on quantization, low precision training, model compression, system modeling and optimizing neural networks for HW acceleration and tooling for neural network eval + debug.

    • Staff Software Engineer, AI Infrastructure
      Jun 2019 - Sep 2022 · 3 yrs 4 mos

      Make AI workloads go fast

  • Engineer at Apple
    Aug 2016 - May 2018 · 1 yr 10 mos

    Helped ship the wireless subsystem for iPhone X and 11.

  • System Engineer at NVIDIA
    Jun 2015 - Sep 2015 · 4 mos

    Power system architecture, found and resolved critical failure scenarios in Nvidia's datacenter gpus.

  • Research Assistant (Integrated Systems Lab) at University of Toronto
    Jan 2014 - Apr 2015 · 1 yr 4 mos

    Novel research in short reach links for high density chip to chip communication.