Professional Experience

Software Engineer - AI

Vega Innovations | Colombo, Sri Lanka

Feb 2025 - Present

At Vega Innovations, I work on cutting-edge AI-driven projects, focusing on autonomous vehicles, quadruped robotics, and AI-powered recognition systems. My work spans deep learning, reinforcement learning, computer vision, sensor fusion, and real-time AI deployments.

  • Vehicle Perception & Sensor Fusion
    • Configured and deployed Autoware on NVIDIA AGX Orin, integrating Hesai XT16 LiDAR via the Nebula package.
    • Developed a monocular depth estimation pipeline using relative and metric depth estimation, improving point cloud generation accuracy for object detection and environment mapping.
    • Engineered a Bird’s Eye View (BEV) system, fusing LiDAR depth data, clustering, speed estimation, and camera pose estimation to enhance the vehicle’s real-time environmental perception.
  • Quadruped – Autonomous Navigation & Perception
    • Designed and implemented a LiDAR-based obstacle detection system, enabling the quadruped to autonomously halt upon detecting immediate obstacles.
    • Developed a custom ROS package for real-time LiDAR data processing, ensuring low-latency obstacle avoidance in dynamic environments.
  • VelloPass – AI-Powered LPR System
    • Conducted research on image enhancement techniques, optimizing vehicle number plate recognition models to improve accuracy across diverse lighting and weather conditions.
    • Contributed to the development and maintenance of a CI/CD pipeline, streamlining deployment, testing, and updates for the AI-driven recognition platform.

AI & Robotics Instructor (Part-time)

ANKA Technologies | Colombo, Sri Lanka

Apr 2025 - Present

Conducting hands-on sessions and workshops on AI, machine learning, and robotics for students and professionals. Teaching practical applications using Python, ROS, and popular AI frameworks.

  • Delivering beginner to intermediate level classes on AI and robotics.
  • Guiding students on projects in computer vision and robot navigation.
  • Introducing practical tools like PyTorch, TensorFlow, and Robot Operating System (ROS).

Associate AI Engineer

Vega Innovations | Colombo, Sri Lanka

Nov 2023 - Feb 2025

As an Associate AI Engineer, I played a key role in developing autonomous systems, reinforcement learning pipelines, and AI-powered applications, with a strong focus on research, real-world implementation, and system optimization.

  • Autonomous Vehicle Project – Multi-Agent & RL-Based Perception
    • Designed and implemented a Multi-Agent System (MAS) for AV simulation, integrating it with reinforcement learning frameworks such as FLAMEGPU2, UniAD, and DaveNet.
    • Developed an End-to-End autonomous driving pipeline, including data collection, preprocessing, training, and evaluation using ResNet, RegNet, and BiFPN architectures.
    • Built a Python module for sensor fusion, integrating LiDAR, RADAR, Camera, GPS, and CAN data, simplifying perception model development.
  • Go2 Quadruped – Navigation, Localization & RL Models
    • Integrated Go2 SDK and ROS2 for real-time motion planning and locomotion control.
    • Implemented SLAM-based localization using Nav2 and LiDAR, optimizing real-time movement strategies.
    • Designed Reinforcement Learning (RL) models for control algorithms, using Isaac Gym for RL training simulations.
  • FaceRec Project – Edge AI Optimization & Anti-Spoofing
    • Optimized a face recognition system for Rockchip Linaro boards, significantly reducing hardware requirements and deployment costs.
    • Developed an RKNN-based face detection pipeline, improving model efficiency for edge AI applications.
    • Integrated depth-sensor-based anti-spoofing techniques, reducing false positives by 70% while enhancing security.
  • VegaGPT – Internal AI Chatbot
    • Developed VegaGPT, an internal knowledge-sharing chatbot for the AI department, leveraging LangChain and OpenAI models to enhance productivity.
  • ChargeNET OS – Custom Linux OS for EV Charging
    • Engineered a lightweight, optimized Linux OS for EV charging platforms, achieving:
      • 400% reduction in memory usage
      • 200% reduction in CPU usage
      • Storage optimization from 64GB to 8GB (portable image reduced to 2GB)

Intern AI Engineer

Vega Innovations | Colombo, Sri Lanka

Jun 2023 - Oct 2023

During my internship at Vega Innovations, I gained hands-on experience in reinforcement learning, computer vision, and AI-driven perception models, working on high-impact projects.

  • Autonomous Vehicle Project – Reinforcement Learning for AV Safety
    • Conducted research on RL algorithms for AV safety, implementing TRPO, PPO, CPO, and PCPO from scratch.
    • Worked with CARLA and Mujoco simulation environments, testing RL models for real-world driving scenarios.
    • Designed safety-critical RL experiments using TorchRL, CleanRL, Omnisafe, and other RL frameworks.
  • FaceRec Project – AI-Powered Attendance System
    • Developed a face recognition system using OpenCV and Kivy, optimized for Raspberry Pi 4B.
    • Enhanced anti-spoofing accuracy to 98% by integrating Intel RealSense depth-based analysis.
  • Bird’s Eye View Project – Semantic Segmentation for AV Perception
    • Conducted research on camera-to-BEV transformations, implementing a UNetXST segmentation model with 85% accuracy.

Mentee

Intel Corporation | Colombo, Sri Lanka

Jun 2023 - Mar 2024

Selected as part of the AWS AI & ML Scholarship Program – Top 500 Cohort, where I gained hands-on experience in AI/ML model development, deployment, and performance optimization under expert mentorship.

  • Mentored by Ali Aram, a Platform Application Engineer at Intel Corporation, receiving guidance on AI/ML engineering best practices.
  • Received career guidance, technical insights, and industry best practices, focusing on aligning academic knowledge with real-world AI/ML applications.

Contributor

Omdena | Colombo, Sri Lanka

Jan 2023 - Mar 2023

Contributed to the Autism Early Detection Project, researching deep learning models for autism diagnosis.

  • Collaborated with Dr. Shaista Hussain on the Autism Early Detection Project, assisting in dataset collection and literature review for deep learning models.
  • Conducted experiments using neuroimaging-based AI models, leveraging ABIDE I & II datasets.
  • Explored deep learning approaches for pattern recognition in fMRI and structural MRI scans.
  • Collaborated with a global AI research team to enhance autism detection accuracy through machine learning techniques.

Founder & Director

heartivVe (PVT) LTD | Colombo, Sri Lanka

May 2019 - Jun 2023

As the Founder & Director of heartivVe, I have led multiple full-stack development, AI integration, and software engineering projects, delivering custom business solutions for clients.

  • Developed 20+ business websites and full-stack solutions, increasing client web traffic by 50%.
  • Built CMS-based learning platforms, streamlining content management and reducing administrative overhead.
  • Designed and developed full-stack applications with Node.js and JavaScript, maintaining 95% test coverage.
  • Optimized project delivery timelines, reducing development cycles by 30% compared to industry benchmarks.