Rachith Prakash

I’m a Research Engineer passionate about building intelligent systems that bridge the gap between simulation and real-world deployment. My work focuses on 3D vision, robotics, and deep learning—designing algorithms, developing simulation frameworks, and creating scalable synthetic data pipelines to accelerate machine learning workflows.

At Intel Labs, I’ve contributed to novel 3D object surface reconstruction methods achieving 90× speedups over traditional approaches, co-developed SPEAR (an open-source photorealistic simulator in Unreal Engine 5), and led automation efforts that cut release times by over 95%. I’ve also driven evaluation pipelines for diffusion models to optimize for real-world deployment. I’m also actively involved in the academic community as part of the organizing committee for the Embodied AI Workshop at CVPR 2025.

My background spans machine learning, computer vision, DevOps, and software engineering, with hands-on experience in C++, Python, PyTorch, Unreal Engine, and cloud infrastructure. I’m always looking for opportunities to solve complex challenges and push the boundaries of AI-driven systems.

I obtained my Masters in Robotics from the University of Maryland, College Park in 2020. During my Masters, I explored Bayesian Deep Learning with primary focus on Reinforcement Learning. I was also a Graduate Teaching Assistant for Dr. Chad Kessens (ENPM662 - Robot Modeling).

I obtained my Bachelors in Electronics and Communications Engineering from the National Institute of Technology Karntaka Surathkal (NITK) in 2016. At NITK, I worked on developing Brain Computer Interface Applications as my final project and was advised by Dr. Deepu Vijayasenan.

Email  /  GitHub  /  Google Scholar  /  LinkedIn

Updates

• [06/2025] I was part of the organizing committee for the Embodied-AI workshop at CVPR '25.
• [07/2024] Generating loads of synthetic data for 3D object's surface reconstruction.
• [06/2024] SPEAR made its 5th open-source release! Check it out here.
• [10/2023] Presented SPEAR at the Intel Labs Open House event.
• [12/2022] SPEAR is now open-source!! 🥳
• [10/2020] Building SPEAR - A Simulator for Photorealistic Embodied AI Research from the ground up!
• [09/2020] I've joined the Intelligent Systems Lab, led by Dr. Vladlen Koltun at Intel, as a research engineer.

Projects

These include coursework, side projects and unpublished research work.

SPEAR and MuJoCo co-simulation

SPEAR and MuJoCo co-simulation early prototype

SPEAR, A Fetch robot doing mobile manipulation

Dijkstra with MMSimulator (C++ implementation)

TSP optmization (C++ implementation)

Coverage Path Planning with Quadcopters!

Traffic sign detection, Belgium Traffic Sign Classification Benchmark dataset

Gazebo Simulation of A-star on Turtlebot3

8 Puzzle problem using Breadth First Search

Motion Planning using Dijkstra Algorithm with obstacle avoidance

Motion Planning using A-star Algorithm with obstacle avoidance

SLAM using factor graphs

Control of Ensembles of Robots with Non-Holonomic Constraints

Canny Edge detector

Adobe Roto Brush (video snapcut)

2-D Panaroma Stitching

Brain Computer Interface Applications

Presentations

A General Framework For Uncertainty Estimation in Deep Learning

Learning Ordinal Relationships for Mid-Level Vision

Credits to Leonid Keselman's website for the source code.