Robotics Engineer - Embedded Systems & Real-Time Vision

ANUJSAI NARAIN

Boston, MA 646-***-**** **********@***.*** Portfolio Linkedin Github

Robotics Engineer with 3+ years of experience and an M.S. in Robotics from NYU. Strong in embedded systems and real-time robotics, spanning perception, navigation, mapping, and controls using ROS2, OpenCV, and TensorFlow. Proven ability to develop deep-learning models, implement visual SLAM, and optimize real-time system performance. Experienced delivering production-quality robotics features from prototyping through deployment with strong ownership, debugging rigor, and cross- functional collaboration.

SKILLS

Languages and OS: Python, C++, Bash, Linux (Ubuntu, Debian, Raspberry Pi OS) Frameworks: ROS2(DDS), OpenCV, TensorFlow, NumPy, Hugging Face Software tools: Git, Docker, CMake, CVAT, STM CubeMX, JSON/XML, YAML Embedded Systems: Raspberry Pi, STM32 boards, ESP32, Arduino, SPI/I2C/UART Integration, RTOS Simulation / Design: Gazebo, RViz, Nav2, MATLAB, Simulink, PLC Ladder, Proteus, SolidWorks, CAD Miscellaneous skills: 3D printing, CNC milling, Soldering, Sensor Integration EXPERIENCE

Robotics Engineering Lead, Rubicon Robotics Inc, New York, USA June 2025 – Present

• Refactored the robotics software framework in ROS2/C++, modularizing perception, control, and DDS pub-sub communication layers to improve scalability and real-time performance.

• Architected hardware and multi-board embedded systems across Raspberry Pi and STM32, developing custom device drivers and integrating IMUs, ultrasonic sensors, temperature sensors, cameras, and encoders via UART/SPI/I C.

• Designed and deployed computer vision pipelines using OpenCV and integrated ML models for detection and motion tracking, coupled with a PID-based control system for autonomous response and real-time data logging to Supabase.

• Improved system reliability through galvanic isolation and a clean-dirty power-domain design, level shifting, ground/noise mitigation, and safe voltage interfacing for mixed-voltage peripherals.

• Implemented an OTA firmware update and custom flashing pipeline for STM32 over UART, incorporating validation and safety checks to enable rapid field iteration without manual reflashing.

• Established CI/CD and testing practices such as regression checks and Test-Driven Development (TDD) for automated builds using Git, Docker and CMake to reduce integration risk and speed iteration.

• Authored setup, build, bring-up, and assembly documentation with clear prerequisites and checklists, standardizing the hardware/software assembly workflow to reduce setup friction and improve repeatability. Robotics Researcher, AI4CE, New York, USA August 2023 – June 2025

• Streamlined MQTT over TCP/UDP for teleoperation on Unitree GO1, ensuring movement and bypassing RPI locks for high-level control, and developed a custom package that reduced latency by over 1 second per instruction.

• Implemented Visual SLAM indoors with LIDAR and omni-directional camera on ROS2 for deployment by integrating pre-mapped data with live input from cameras, debugging and optimizing real-time perception pipelines.

• Collaborated the integration of Husky mobile robot by Clearpath Robotics with xArm industrial robotic arm, implemented real-time control with ROS2 and RTOS concepts for object manipulation, essentially to open doors. Computer Vision Engineer, Building Diagnostics Robotics, Inc., New York, USA September 2022 - July 2023

• Trained a deep-learning model to accurately detect multiple defects on sets of roughly 25 datasets of 100 images each, including thermal, air leakage, and presence of moisture on building floors using thermal and colored images.

• Collaborated with a team of 5 engineers to develop a Docker container that increased model precision by 15%. Implemented transfer learning techniques to fine-tune model architecture for defect detection.

• Created a comprehensive catalog detailing visual attributes present in thermal images, thereby establishing a Standardized Operating Procedure and reference guide to ensure precise labelling for training procedures. PROJECTS

RoomieAI – The Home Assistant

• Designed and implemented a robotic system simulation in ROS2 using Gazebo plugins, CMake, and XML files for simulation. Integrated SLAM with LIDAR for autonomous navigation, allowing the bot to navigate accurately.

• Introduced object detection algorithms for localization for inventory management. Leveraging LLM capabilities, prompt engineering and speech recognition to further automate and optimize the process, enhancing IoT operations.

• Conducted a comparative analysis to assess the functionality of various LLM models within the robot system, including LLaMa 2-7B, LLaMa2-13B, and Oracle 3B GGUF models, optimizing performance by 10%. EDUCATION

New York University Master of Science in Mechatronics and Robotics CGPA: 3.84 August 2022 – May 2024 Manipal Institute of Technology Bachelor of Technology in Mechatronics August 2016 - May 2020

Contact this candidate