> For the complete documentation index, see [llms.txt](https://hasithz.gitbook.io/hasith-abayakoon/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://hasithz.gitbook.io/hasith-abayakoon/readme.md). # Hasith Abayakoon {% hint style="info" %} Tech-Savvy AI Engineer with a record of designing and implementing deep learning solutions that improve user experiences and operations. Committed to lifelong learning and adopting new technologies to stay at the forefront of the AI field. {% endhint %} ***
Projects/pages/LM3u5GzkXUT0nPICvTIS/files/KOYEsZLvN4BdIQYd5PcH
Blogs/pages/biT78um1qaJiVDvZ8qKl/files/HheSf56gqe0X6IMn4qMd
Certificates /pages/EgIkGUrnwvhL6V4pSEVT/files/Ki3QzaIJDdH6kQ1T6GT9
Open Source Contributions/pages/pUwgp3jfA90Eas3l8MiY/files/GYl01r23aVVY5c4nNDFz
Pearson Contributions/pages/4Vn21UNKDtDLRQva5LFQ/files/soWwkCDn14xlWdZoyGwu
Github/pages/SA3Yu1aDvceFCEUEieaj/files/lxQaQa3ZWbSVnrhCnXfH
*** ## Contact | +94 7666 17 496 | | | ------------------------- | ----------------------------------------------------------------- | | :golf: Colombo, Sri Lanka | :link: [linked-In](https://www.linkedin.com/in/hasith-abayakoon/) | *** ## Experience
Senior Software Engineer - Pearson (2024-05 / Present) * Creating gitlab pieplines for automate all CICD. *** #### Skills ``` Python / Gitlab / Docker / AWS / CICD ```
Research Lead - EagleEye - (2023-05 / 2024-05) * Spearheaded the development of the company’s flagship AI project, guiding it from concept to deployment. * Architecture the core AI systems, ensuring scalability, reliability, and performance. * Fostered innovation by staying abreast of industry trends and emerging technologies in AI and machine learning. *** #### Skills ``` Python / PyTorch / Docker / Kafka / Celery / ViT / GNN,GCN ```
Research Engineer – Technology Consultant - BooleanLabs - (2023 -05/ 2024-05) * Created and fine-tuned cutting-edge AI models, including Convolutional Neural Networks (CNNs) and Graph Convolutional Networks (GCNs), to solve complex problems and enhance analytical capabilities. * Conducted extensive experimentation with various Vision Transformer (ViT) models, optimizing them for superior performance in image recognition tasks. * Kept abreast of technological advancements in the AI field, integrating innovative practices and tools to drive continuous innovation and growth. *** #### Skills ``` Python / PyTorch / Docker / ViT / GNN, GCN ```
Research Assistant - Sri Lanka Institute of Information Technology - (02/2022 – 04/2023) * Engineered robust architectures for robotic applications, integrating advanced deep learning and reinforcement learning pipelines to enhance autonomy and efficiency. * Innovated and programmed custom virtual environments tailored for reinforcement learning, enabling simulation and training of AI models in controlled, scalable settings. * Designed and implemented a data transfer system between Python 2.7 and 3.x, utilizing sockets and multi- threading to ensure seamless, real-time data communication. * Developed and deployed TensorBoard and Streamlit visualization tools to monitor and interpret deep learning model metrics, facilitating data-driven decisions and model optimizations. * Advanced knowledge and practical experience in deep learning frameworks such as PyTorch, TensorFlow, and Keras, applying them to create and iterate on state-of-the-art AI models. *** #### Skills ``` Reinforcement Learning / PyTorch / Python 2.X / Python / NAO Robot ```
Python Software Developer - Orange Electric Panaluwa - (09-2021 – 09-2022) * Leveraged expertise in computer vision technologies, including OpenCV, Google Cloud Vision API, and scikit- image, to develop and enhance image analysis applications. * Designed and integrated RESTful APIs to facilitate seamless front-end to back-end data exchange and application functionality. * Deployed sophisticated deep learning models on edge devices like Raspberry Pi and NVIDIA Jetson, optimizing for high performance in resource-constrained environments. * Proficient in developing scalable web applications using Python frameworks such as Django and Flask, adhering to best practices and modern web standards. *** #### Skills ``` OpenCV / Nvidia Jetson / Django / SK learn ```
Electronic and Electrical Engineer Trainee - JLanka Technologies - (2018 – 2019) * Conducted comprehensive solar panel and string testing to ensure efficiency, reliability, and compliance with industry standards. * Performed rigorous optimizer tests to maximize energy output and system performance, implementing enhance- ments to improve solar yield. * Configured and calibrated inverters, optimizing settings for improved energy conversion and grid compliance, contributing to overall system efficacy and sustainability. *** #### Skills ``` solar panel / optimizer tests ```
*** ## Publications
Reinforcement Learning Approach for Dialogue Management to Aid Child-Robot Interaction in ASD Screening -*1st #### 2023 6th International Conference on Artificial Intelligence and Big Data (ICAIBD) *** 2022 - Published :link: paper *** This paper investigates how Q-learning, State Action Reward State Action (SARSA), and Deep Q Networks (DQN) algorithms in Reinforcement Learning (RL) are used for dialogue management in a child-robot interaction system. Based on observations of the child's behavior and response to questions, the system determines how to structure the next question to maintain the child's engagement in a dialogue that is developed based on Autism Diagnostic Observation Schedule (ADOS). To accomplish this, a virtual environment is created to simulate interactions between the child and the robot, allowing for the development of an initial model that can be refined and tested in the real world. This paper presents how robust a DQN is to withstand an increasing variety of questions in dialogues. Further, it has been found SARSA is the fastest converging algorithm to learn the optimal policy for such interactions.
Child–Robot Interaction in Robot-Enhanced Diagnostic and Therapeutic Sessions of Children -*1st #### **40th National Information Technology Conference (NITC)** *** 2022 - Published :link: [paper](https://www.researchgate.net/publication/369451993_Child-Robot_Interaction_in_Robot-Enhanced_Diagnostic_and_Therapeutic_Sessions_of_Children) *** Child-Human-Robot interaction might be more authentic, intuitive, successful, and natural if robots had emotional intelligence. Therefore, observing and recognising human actions is vital for achieving this goal. As well as that, making robots appealing to children is similarly crucial. Otherwise, children may be intimidated by humanoid robot agents, which would hinder child-robot interaction. Through the use of a robotic agent, the NAO robot, we determine child-robot interaction based on the degree of the child's capability of keeping the gaze with the robotic agent during child diagnostic and therapeutic sessions. The Wizard of OZ (WoZ) approach is used to remotely control the robotic agent in Robot Enhanced Therapy (RET), where the child's guardian or therapist acts as the third party and the robot is controlled by a different individual. Based on the findings, we suggest cutting-edge research to effectively establish child-robot interaction in robot-assisted and robot-enhanced therapeutic sessions for children. In addition, we provide crucial insights for enhancing child-robot interactions in order to diagnose neurodevelopmental disorders in children and facilitate future research.
Child–Robot Interaction: Evaluation of Child Attention in Real-time with NAO Robot -*1st #### IEEE Region 10 Conference (TENCON 2022) *** 2022 - Accepted :link: paper *** Abstract—Based on four different tutoring events, this paper presents a comprehensive evaluation of the attentiveness of children to numerous instructions given by a robot. Throughout each session, the NAO-humanoid robot served as the instructor. Index Terms—Child Robot Interaction, Child Attention Eval- uation, NAO
Distance Measuring Device to Calculate the Height and Width of the Human Ear Lobe *1st #### Wayaba University of Sri Lanka (Medical Faculty) *** 2022 - Published :link: [paper](https://www.researchgate.net/publication/367464784_Distance_Measuring_Device_to_Calculate_the_Height_and_Width_of_the_Human_Ear_Lobe) *** Unique morphometric characteristics of human ear lobes are widely used in recognizing and identifying individuals. The currently employed traditional anthropometric instruments take more time and demand direct touch with the subject. Health care instructions to maintain physical distances which limit close contact with the individual is a major obstacle in morphological and morphometric analysis of human ear. Therefore, the aim of this project was to develop a device that can be mounted on any smart mobile phone and allows the user to measure the height and width of the external ear without touching the individual itself. For this project, a laser-based time-of-flight (TOF) camera; a range imaging camera system employing TOF techniques to resolve the distance between the camera and the subject for each point of the image, by measuring the round-trip time of an artificial light signal provided by a laser or a light emitting diode (LED) was used. These TOF cameras are also a part of a broader class of scanner-less Light Detection and Ranging (LIDAR), in which the entire scene is captured with each laser pulse, as opposed to point-by-point with a laser beam such as in scanning LIDAR systems. During the process of designing this device, two TOF sensors were used to develop the module and attached to a device where it can be measured and adjust the distance of the two sensors. Two lasers were used to identify the points of the surface and to get the convenient position of the surface that needed to be measured. Then the camera sensor was used in the smartphone provided to capture the image. The camera sensor was accessed using the droid cam extension and transmitted through Wireless Fidelity (Wi-Fi) to a computer in order to view the frame. The frames were acquired using the OpenCV library and computed the pixcl values using the NumPy library. The distances were drawn using the line function providcd in the library. When using this device, it requires a separate custom-made soliware where it 1akes values wirclessly (Wi-Fi) and where he custom-made device sends the data to the personal compuler (PC) custom-made sofiware using python programnming language where connects with a smartphonc camera to acquire the image. For the smartphone the droid Cam application is one of the examples we can use to act as a remote camera. Once an image is taken, the image will be preprocessed and displayed in the software with the measurements. Then the two measuring devices, one to align the camera module with the test subject and one to measure the distance from the test subject will be used to take the final measurement with an accuracy of Imm. In conclusion, this device can be used for anthropometric measurement of human ear lobes with an accuracy of Imm and without having close contact with the subjected individual. Further, as this allows the user to measure the width and height of a flat surface using vision recognition, the device can be repurposed according to the necds of the project. The programs can be added to the embedded device using various devices. Keywords: Human Ear Lobe; Measuring Device; OpenCV; ToF; Vision
*** ## Education
Master of Business Admintration (Marketing) University :
BSc Engineering (Hons) in Electrical & Electronic Engineering University :
*** ## Volunteering
AIESEC - IGV Manager - (2018 - 2020) * **Number of Volunteers Recruited**: Measures the total number of international volunteers recruited for various projects. This reflects your effectiveness in attracting volunteers. * **Volunteer Satisfaction Score**: Gauges the overall satisfaction of volunteers with their experience. This can be measured through surveys or feedback forms. * **Project Completion Rate**: Tracks the percentage of projects that are completed successfully with the help of volunteers. This indicates the effectiveness of volunteer contribution towards projects. * **Host Organization Satisfaction**: Measures the satisfaction of the organizations or communities hosting the volunteers. Like volunteer satisfaction, this can be assessed through surveys or direct feedback. * **Cultural Exchange Effectiveness**: Evaluates how well the program facilitates cultural exchange and learning. This can be subjective but is key to AIESEC's mission of cross-cultural understanding. * **Volunteer Retention Rate**: Looks at the percentage of volunteers who choose to engage in another AIESEC opportunity. High retention rates can indicate a positive overall experience. * **Time to Match Volunteers with Projects**: Tracks the average time it takes to match volunteers with suitable projects. Efficiency in this process is crucial for a smooth operation. * **Diversity of Nationalities**: Measures the diversity of the volunteer pool in terms of nationalities. A diverse group suggests a wide-reaching appeal and effectiveness in global engagement. * **Social Impact Measurement**: Assesses the social impact of the volunteer projects, which can be quantified through various metrics depending on the project's nature and goals. * **Compliance with Legal and Safety Standards**: Ensures all activities comply with local and international legal and safety standards, which is crucial for organizational reputation and volunteer safety. * **Alumni Engagement**: Tracks the level of engagement of past volunteers, which can be a resource for mentorship, networking, and community building.
***