Manoj Kumar Selvaraj

Embedded | Firmware | Electrical | Robotics

Manoj Kumar Selvaraj

About Me

Hello, I'm Manoj Kumar Selvaraj – a passionate Embedded Systems Developer and Robotics Engineer based in Maryland, USA. With an M.Eng. in Robotics from the University of Maryland and a B.E. in E.C.E from PSG Institute of Technology, I work across the full embedded stack - from hardware design, sensor fusion, and board-level bring-up to firmware development, RTOS integration, edge AI optimization, and FPGA acceleration. I enjoy designing robust hardware–software systems, validating high-reliability platforms, and building intelligent wearable and edge devices that operate with precision in real-time.

Skills

Programming

Tools & Frameworks

Protocols

Hardware

Education

Jan 2024 - Dec 2025

University of Maryland College Park

Masters in Robotics

Jun 2017 - Apr 2021

PSG Institute of Technology and Applied Research

B.E. in Electronics and Communication Engineering

Work Experience

May 2025 - Present

37degrees, Inc, Chicago

Lead Embedded & Firmware Development Intern

  • Led full lifecycle embedded development for the CultureON portable incubator, taking the system from early concept through EVT → DVT → PVT and delivering pilot-ready hardware and firmware.
  • Architected the hardware platform, including sensor/actuator selection, power-tree design, protection circuitry, and interfaces; designed 5 custom PCBs and assembled five production-quality units.
  • Integrated the TPS25947 power switch to replace diode-based muxing, improving overall power efficiency by ~15% for continuous-duty operation.
  • Developed production-grade firmware using FreeRTOS, implementing multi-task scheduling, inter-task messaging, and sensor fusion with deterministic control of actuators.
  • Implemented the connectivity stack, including custom BLE GATT services for mobile/desktop integration and Wi-Fi station mode + MQTT telemetry for secure cloud-based data streaming.
  • Built a reliable OTA update pipeline with image validation, rollback capability, and version control for remote firmware deployment.
  • Developed manufacturing test firmware and automation scripts, reducing board bring-up and verification time from ~2 hrs → ~30 mins per unit.
May 2021 - Dec 2023

YantraVision Software Pvt Ltd, Bangalore

Senior Software Engineer

Client [Xilinx-AMD] : Multicore DSP Kernel Optimization & Validation (Hardware Accelerator, HPC, FPGA)

  • Accelerated and benchmarked DSP libraries (GeMM, 2D-FFT) on FPGA (Versal ACAP) and AIE SIMD processors using Xilinx Vitis, documenting detailed performance metrics.
  • Optimized a 4096-point FFT on AI Engine by reducing data-transfer bottlenecks, achieving 1.5× MSPS speedup over the baseline HLS kernel.
  • Established a Jenkins-based CI/CD regression suite, automating functional and performance testing for 72+ DSP libraries.
  • Designed and validated single- and multi-core kernels for AIE; performed pre-silicon validation (QEMU) and post-silicon bring-up including power characterization.
  • Spearheaded a three-member team to develop a multithreaded host application on ARM Cortex-A72, reducing 5G FR2 signal-processing latency by 75%.
  • Integrated Linux kernel driver interfaces (XOCL) with XRT APIs for efficient interrupt handling and inter-process communication between the host and FPGA.

Machine Vision Systems & High-speed Imaging for Carton Printing (Computer Vision, Real-Time Processing)

  • Implemented GPU acceleration using CUDA and OpenMP, maintaining inspection throughput at 300–350 meters per minute while improving computational efficiency.
  • Engineered automated workflows for annotating 40,000+ Braille points on medical packaging, achieving 94.8% detection accuracy at 50 FPS.
  • Enhanced Braille inspection speed using Haar Cascade classifiers, achieving ~5 ms processing time per carton with improved defect-detection reliability.
  • Developed robust image-processing pipelines using dark-field illumination to detect embossing defects and ensure consistent print-quality inspection.
Mar 2021 - May 2021

Zoho Corporation Pvt Ltd, Chennai

Project Trainee

  • Contributed to backend development of the Zoho "Cliq" application, resolving issues and optimizing functionality.
  • Diagnosed and resolved critical backend bugs, improving application stability and reducing downtime.

Projects

Autonomous Mobile Robot Development

Technologies: Sensor Fusion, OpenCV, Real-Time Control
Timeline: Feb 2025 - May 2025
An embedded control system was developed to drive an autonomous mobile robot capable of navigating a structured arena, detecting color-coded blocks, and completing a full pick-and-place mission.

View Code

On-Device Gait Prediction for Embedded Wearables

Technologies: Wearable IMU, TFLite, TinyML
Timeline: Oct 2025 - Dec 2025
A real-time, on-device gait prediction system that processes multi-axis IMU signals and runs fully quantized neural networks on ultra-low-power microcontrollers.

View Code

SRAM based Compute-In-Memory Accelerator for Neural Network

Technologies: Mixed Signal Design, Analog Circuits, Cadence
Timeline: Oct 2025 - Dec 2025
Designed a 5.2kb SRAM-based AI accelerator with 80-bit parallel throughput, implementing custom control logic and a fixed-point data pipeline. Verified system performance in Cadence, achieving 90.9% classification accuracy on MNIST.

View Code

TinyML Speech Command Classification

Technologies: CNNs, TFLite Micro, Embedded AI
Timeline: Sep 2025 - Oct 2025
An embedded deep-learning pipeline for real-time speech command recognition, featuring quantized CNN deployment on a Cortex-M4F microcontroller.

View Code

Perception-Driven TurtleBot Navigation

Technologies: ROS2, OpenCV, YOLO
Timeline: Apr 2024 - May 2024
Integrated an autonomous system enabling path following, real-time stop sign detection, and dynamic obstacle detection using computer vision and machine learning algorithms.

View Code

Intelligent Parking for ADAS

Technologies: BLE, Android App, Firebase
Timeline: Jan 2021 - Jun 2021
Led a team of four to develop an indoor Bluetooth-based localization system to detect empty parking spots and navigate users to available spaces within a building.

View Code

Smart Energy Meter

Technologies: Embedded C, MQTT, Sensors
Timeline: Jan 2019 - Feb 2019
A microcontroller-driven smart energy meter that performs on-device signal acquisition, power computation, and cloud reporting through a WiFi-enabled firmware stack.

View Code

Dynamic Obstacle Avoidance using RL Algorithms

Technologies: Q-Learning, DQN, Python
Timeline: Feb 2024 - May 2024
Implemented Q-Learning and Deep Q-Learning for dynamic obstacle avoidance using ray-sensor perception and reinforcement learning.

View Code

Comparison of Path Planning Algorithms

Technologies: RRT, RRT-Connect
Timeline: Jan 2024 - Apr 2024
A comparative study of RRT, RRT-Connect, and Improved RRT-Connect for robotic path planning, evaluating execution time, path quality, and overall efficiency.

View Code

Achievements & Leadership

Workshop Facilitator

Convened a two-day workshop on "Embedded systems - PIC Microcontroller" at PSG iTech, engaging 100+ students.

Published Research

Published a paper titled "Intelligent Parking for ADAS" at NC2SD 2021.

Best Project Award

Won Best Project Award for "Smart Energy Meter".

Contact

Feel free to reach out.

Location

Maryland, USA