Josyula Krishna profile photo 1 on Tal

Josyula Krishna

Robotics Software Engineer

Hyderabad

4+ years of work experience

Perceptyne
Apr 20251 yr 5 mosPresent

Hyderabad, Telangana, India

Bringing Physical AI to humanoids to tackle industrial and household tasks.
Dexmate
Oct 20246 mosMar 2025

Santa Clara County, California, United States

Robotic Manipulation for semi-humanoid to pick and place totes.
Virginia Tech Transportation Institute
Jun 20231 yr 5 mosOct 2024

United States

• Designed and Developed a customized CARLA based Simulation environment for Autonomous Truck Navigation in the Port of Virginia integrated with ROS. • Engineered a LiDAR-based 3D object detection system for a GPS-denied visual navigation platform and seamlessly integrated it with a PID controller. • Developed a sparse point cloud 3D fence detection algorithm using PV-RCNN achieving 98% accuracy and successfully integrated into ROS using OpenCV CuDNN ONNX capabilities, enhancing real-time reverse path planning for an autonomous truck. • Coded a novel multi-threaded LiDAR based curb-detection module in PCL C++ library. • Integrated ADAS ROS modules in C++ for testing collision avoidance system framework on smart road, which allows to test last-minute collision avoidance strategies by introducing obstacles in front of a moving vehicle. • Authored test cases for ADAS behavior in the presence of static targets on roads. • Worked with interfacing various hardware components such as braking actuators, LiDARs, steering motors for establishing communication with onboard embedded hardware in C++ and Python. • Dockerized the code for seamless execution on platforms utilizing Ubuntu low-latency kernels. • Implemented software which runs on Git CI/CD pipelines, enhancing software deployment processes and ensuring continuous integration and delivery. • Developed a Time-to-Collision (TTC) based collision avoidance system for an autonomous truck utilizing LiDAR-based object detection.
Fujitsu
Jun 20224 mosSep 2022

• Research focused on Reinforcement Learning for Combinatorial Optimization problems with a direct application to traffic flow management • Formulated RL-based strategies for Min k-cut and Maxcut using Graph Neural Networks, and Pointer Networks, demonstrated results on Graphs with 40 nodes achieving Mincut with constant edge weights.
Oregon State University
Dec 20211 yr 7 mosJun 2023

Corvallis, Oregon, United States

• Developed an accurate 3D volumetric reconstruction using methods from 3D computer graphics for cherry trees the mesh obtained is used for simulated virtual training of robots for pruning. This method produces 98% accurate measurements of tree width estimates and eliminates manual inspection of tree measurements • Synthetic data generation for tree anatomy detection, segmentation and foreground background disambiguation using from GAN and FlowNet based deep learning architectures for accurate foreground and background extraction in Vineyard Pruning with UR5 robot arm. • Developed the first fully vision based pruning system using MaskRCNN and Mask2Former architecture for 99% accurate detection of tree anatomy and pruning cherry trees which has been deployed and tested at Olson Farms to produce 58% accurate cuts. https://github.com/josyulakrishna/detectron2_pruning
GitHub
@josyulakrishna

Josyula’s AI Stack

codex
codex
cursor
cursor

Education

Oregon State University

Master of Science - MS, Mechatronics, Robotics, and Automation Engineering

Oregon State University

20212 yrs2023
International Institute of Information Technology Hyderabad (IIITH)

Graduate Studies In Robotics, Mechatronics, Robotics, and Automation Engineering

International Institute of Information Technology Hyderabad (IIITH)

20165 yrs2021

Join Josyula and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Josyula Krishna on Tal

Headline

Robotics Software Engineer

About

Robotics Engineer with 7+ years of experience in autonomous systems, computer vision, and AI. Expert in developing advanced navigation algorithms, LiDAR-based 3D Object detection, and multi-robot collision avoidance systems for autonomous vehicles and drones. Proven track record in agricultural robotics, industrial automation, and R&D projects for Collins Aerospace, Intel, and Fujitsu. Skilled in bridging research with practical implementation using C++, ROS, and machine learning techniques.

Location

Hyderabad

AI stack

codex

cursor

Work experience

AI/ML Engineer at Perceptyne, Apr 2025 - Present · 1 yr 5 mos, Hyderabad, Telangana, India, Bringing Physical AI to humanoids to tackle industrial and household tasks.

Robotics Engineer at Dexmate, Oct 2024 - Mar 2025 · 6 mos, Santa Clara County, California, United States, Robotic Manipulation for semi-humanoid to pick and place totes.

Robotics Engineer at Virginia Tech Transportation Institute, Jun 2023 - Oct 2024 · 1 yr 5 mos, United States, • Designed and Developed a customized CARLA based Simulation environment for Autonomous Truck Navigation in the Port of Virginia integrated with ROS. • Engineered a LiDAR-based 3D object detection system for a GPS-denied visual navigation platform and seamlessly integrated it with a PID controller. • Developed a sparse point cloud 3D fence detection algorithm using PV-RCNN achieving 98% accuracy and successfully integrated into ROS using OpenCV CuDNN ONNX capabilities, enhancing real-time reverse path planning for an autonomous truck. • Coded a novel multi-threaded LiDAR based curb-detection module in PCL C++ library. • Integrated ADAS ROS modules in C++ for testing collision avoidance system framework on smart road, which allows to test last-minute collision avoidance strategies by introducing obstacles in front of a moving vehicle. • Authored test cases for ADAS behavior in the presence of static targets on roads. • Worked with interfacing various hardware components such as braking actuators, LiDARs, steering motors for establishing communication with onboard embedded hardware in C++ and Python. • Dockerized the code for seamless execution on platforms utilizing Ubuntu low-latency kernels. • Implemented software which runs on Git CI/CD pipelines, enhancing software deployment processes and ensuring continuous integration and delivery. • Developed a Time-to-Collision (TTC) based collision avoidance system for an autonomous truck utilizing LiDAR-based object detection.

Research Intern at Fujitsu, Jun 2022 - Sep 2022 · 4 mos, • Research focused on Reinforcement Learning for Combinatorial Optimization problems with a direct application to traffic flow management • Formulated RL-based strategies for Min k-cut and Maxcut using Graph Neural Networks, and Pointer Networks, demonstrated results on Graphs with 40 nodes achieving Mincut with constant edge weights.

Graduate Research Assistant at Oregon State University, Dec 2021 - Jun 2023 · 1 yr 7 mos, Corvallis, Oregon, United States, • Developed an accurate 3D volumetric reconstruction using methods from 3D computer graphics for cherry trees the mesh obtained is used for simulated virtual training of robots for pruning. This method produces 98% accurate measurements of tree width estimates and eliminates manual inspection of tree measurements • Synthetic data generation for tree anatomy detection, segmentation and foreground background disambiguation using from GAN and FlowNet based deep learning architectures for accurate foreground and background extraction in Vineyard Pruning with UR5 robot arm. • Developed the first fully vision based pruning system using MaskRCNN and Mask2Former architecture for 99% accurate detection of tree anatomy and pruning cherry trees which has been deployed and tested at Olson Farms to produce 58% accurate cuts. https://github.com/josyulakrishna/detectron2_pruning

Education

Master of Science - MS, Mechatronics, Robotics, and Automation Engineering, Oregon State University

Graduate Studies In Robotics, Mechatronics, Robotics, and Automation Engineering, International Institute of Information Technology Hyderabad (IIITH)

Public links

github: https://github.com/josyulakrishna