Engineer / Local AI / Robotics

PiyushMedikeri

I build and experiment with local AI and robotics, exploring new models and understanding the hardware behind them. Away from the keyboard, I enjoy dancing, table tennis, and hiking.

Explore my experience
Santa Clara, California

01 Career & capabilities

Professional
experience.

Engineering roles and internships at NVIDIA, MathWorks, PLEN Robotics, and Giscle India, spanning AI performance, GPU-accelerated robotics, developer tools, and technical enablement.

Roles & responsibilities

  1. NVIDIA to Present

    1. Sr. System Software Engineer, AI & Robotics

      to Present

    2. System Software Engineer, AI & Robotics

      to

    Local AI, Robotics, Developer Enablement & Technical Marketing

    • Evaluate AI and robotics products for usability, performance, and documentation; provide developer-focused feedback.
    • Build and present demos for Jetson and local AI, covering speculative decoding, open models, and agent workflows.
    • Deliver Jetson and DGX Spark workshops, DLI courses, and livestreams on practical AI workflows.
    • Help reviewers evaluate products through reproducible benchmarks, technical guides, and hands-on support.
    • Build hardware-access and inventory tools, plus automated benchmarks and model/agent accuracy evaluations.
  2. MathWorks to

    Remote · Natick, Massachusetts

    ROS Developer Intern

    Simulink · ROS Toolbox Team

    • Built a Simulink block for ROS Toolbox to time-align data from multiple sensors, supporting synchronized inputs in ROS workflows.
    • Developed a harness to measure ROS throughput and analyze performance, translating results into tuning guidelines adopted by the team.
  3. PLEN Robotics to

    Osaka, Japan

    Robotics Intern

    PLEN Cube applications & product demonstrations

    • Helped refactor the Python codebase and integrate ROS-based functionality into applications running on PLEN Cube.
    • Helped build coffee ordering with face recognition and order handling, alongside a room-checking application for PLEN Cube.
    • Participated in product introductions to other companies, presenting PLEN Cube and demonstrating its capabilities through application examples.
  4. Giscle India2017 · 3-month internship

    Bengaluru, India

    Intern

    Self-driving toolkit & driver monitoring

    • Contributed to a self-driving-car toolkit, including processor-related work.
    • Built a driver-drowsiness prototype with an Intel RealSense camera and an alarm triggered after a configurable eye-closure delay, initially five seconds.

    Intel RealSense · Computer vision · Prototyping

02 About me

A little
about me.

Santa Clara, California

Piyush beside an NVIDIA inference demonstration at CES 2026.
01 / At the demo deskCES 2026

I like building things, bringing people together, and trying something new.

I enjoy turning an idea into something real, especially with people who bring a different perspective. I’m comfortable taking the lead, helping a group find its direction, and making sure everyone has a chance to contribute. I take what I do seriously, but there’s always room for a laugh along the way.

Outside work, I love dancing, a good game of table tennis, and getting out for a hike. I’m always up for a little friendly competition or a walk through a different downtown. I enjoy a change of scene, good company, and discovering somewhere new.

03 Talks, demos & community

Public
presence.

I enjoy bringing people into the conversation: sharing what I know, learning from others, and helping someone try something new.

Piyush delivering a DLI session at an NVIDIA GTC podium.
02 / Delivering a DLI sessionJetson DLI session

Teaching · DLI sessions

Jetson & DGX Spark

I deliver hands-on sessions on Jetson and DGX Spark. At GTC 2026, I was a lab instructor for local AI setup, inference, and agent workflows on DGX Spark.

Explore the GTC 2026 lab
A view from the front of a workshop, with participants working at rows of laptops.
03 / In the roomHands-on learning

Open course · Co-instructor

Isaac Sim + Jetson

I co-taught an Isaac Sim + Jetson course on connecting simulated robots with real hardware, and helped publish the course materials openly.

Explore the open course
Piyush and an attendee looking at a demonstration on a monitor.
04 / At the demoBuild A Claw event

Live demos

Build A Claw

At Build A Claw, I shared a working demo and explored the technology with attendees. I enjoy the questions and ideas that come from trying things together.

Explore the Build A Claw event
Example TreeHacks event artwork featuring Stanford’s tower and green mountains, dated February 14–16, 2025.
TreeHacksExample event artwork

Hackathons & community

As a builder. And a judge.

I don’t just build at hackathons. I’ve also served as a judge at TreeHacks and other hackathons, reviewing projects and learning from other builders.

I worked on Robot navigation using VLMs in Isaac Sim, which earned third place at an NVIDIA hackathon. It used a vision-language model, Isaac Sim, and Jetson Orin to interpret camera input and guide a simulated robot’s navigation.

See the navigation demo

04 Projects

Projects &
experiments.

A selection of software tools, local AI experiments, and earlier robotics prototypes.

Autonomous harvesting robot with a Raspberry Pi, camera, and gripper.

Computer vision

A robot that picks fruit

An E-Yantra project that used a camera to find the requested fruit, pick it with a gripper, and deliver it to the right place.

Staircase-climbing wheelchair prototype with a robotic arm.

Assistive robotics

A stair-climbing wheelchair

An Arduino-based prototype built in 24 hours, using a gyroscope for stability. It placed first among 30 national teams and later received an IEEE Best Hackathon Project award.

Device Pool dashboard showing device availability, reservations, operating systems, and hardware-health metrics.

Developer tools

Device Pool

A platform I built for finding, reserving, and connecting to shared Linux, macOS, and Windows development machines, with device-health monitoring.

  • Next.js
  • TypeScript
  • PostgreSQL
  • Docker
See the code
V-REP simulation showing striker and goalkeeper reinforcement-learning agents.

Reinforcement learning

Learning to score and save

A simulated striker and goalkeeper trained against each other using Q-learning in V-REP. An experiment in how actions and rewards shape behavior.

Front view of the NVIDIA DGX Spark used for local AI experiments.

Personal project

Local AI at home

A DGX Spark setup for trying models and running a Qwen-powered research and email service, with scheduling, duplicate checks, and recovery when services fail.

  • Qwen
  • DGX Spark
  • Agent workflows
Package-delivery routing and planning simulation in ROS and Gazebo.

Planning

Planning a delivery route

A ROS and Gazebo project comparing two ways to group packages and plan delivery routes, breaking one large routing problem into smaller ones.

Illustrative image of an LED cube glowing blue and purple.

Creative electronics

LEDs synced with music

I was bored, so I built LED lights that sync with music beats. A small, just-for-fun experiment.

Watch the LED demo
Illustrative image of a fatigued driver resting on a steering wheel.

Computer vision · Giscle India

Sleep Warning

Built during my Giscle India internship, this prototype used an Intel RealSense camera to detect prolonged eye closure and sound an alarm after an adjustable delay, initially five seconds. It was also tested with an Ola driver.

Explore the Giscle prototype

05 Off the clock

Outside
of work.

Hiking, table tennis, dancing, and saying yes to something new.

Piyush hiking on a wooden mountain trail surrounded by greenery, snow patches, and mist.
05 / Out on a trailHiking

A little adventure. A lot of fun.

Always up
for something.

I love a good hike, a competitive game of table tennis, and any excuse to dance. I’m also trying skiing and snowboarding. There’s always something new to learn.

And I love a good party. Great music, good company, and a reason to celebrate: I’m very much a “work hard, party harder” person.

A group selfie on a snowy ski slope, with helmets and goggles.
06 / A snow dayOn the slopes

06 Get in touch

Say hello.

Have a project in mind, an interesting engineering role, or a shared interest? Let’s talk.

piyushmedikeri1997@gmail.com
LinkedIn ↗ GitHub ↗

A few frames from my world.