
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
AI Research Scientist @ ASAPP
With over 6 years of experience at ASAPP, a leading startup in conversational AI and natural language processing, I am currently an AI Research Scientist who creates and improves cutting-edge solutions to solve problems in the customer service domain. I believe my diverse background in research, product management, engineering and solutions architecture makes me a versatile team player. My passion is to apply my skills in product development, machine learning research, team leadership and engineering to work with teams in creating innovative and impactful products, that will help solve problems and enhance customer experiences. I am always eager to learn new things and challenge myself with complex and novel problems. My goal is to contribute to the advancement of artificial intelligence and make a positive difference in the world.
University of Alberta
Certificate, Software Product Management
January 1, 2020 – January 1, 2020
University of Alberta
Certificate, Software Processes and Agile Practices
January 1, 2020 – January 1, 2020
CIMA
Advanced Diploma in Management Accounting
November 1, 2013 – Present
Curtin University
Bachelor’s Degree, Computer Systems Engineering
January 1, 2011 – January 1, 2015
ASAPP
Product Manager (Rotation) - Intents and Coaching Insights
May 1, 2022 – August 1, 2022
Hybrid
ASAPP
AI Research Scientist / Research Product Manager
February 1, 2022 – Present
Hybrid
ASAPP
Machine Learning Research Engineer
October 1, 2018 – February 1, 2022
Hybrid
WSO2
Solutions Engineer
August 1, 2017 – October 1, 2018
Colombo, Sri Lanka
LSEG Technology
Software Engineer (Machine Learning)
November 1, 2013 – July 1, 2017
Malabe, Western Province, Sri Lanka · On-site
Tekkit
Director Of Technology
March 1, 2012 – February 1, 2015
Colombo, Sri Lanka
Breast Cancer Detection and Classification
October 1, 2016 – Present
Trying to develop a full pipeline that can detect and classify breast cancer using mammogram analysis. While we are still researching on this topic, the expected final outcome is a system that takes breast mammograms as inputs and alerts if the mammogram shows abnormality and also whether it’s benign or malignant.
Domain Specific Language Development (DSL)
December 1, 2015 – December 1, 2016
This DSL is at an abstraction level close to the business specification, and uses deterministic inference to allow system definitions that are as much as 500 times more compact than the corresponding code. This revolutionizes the system engineering process by radically reducing development time, improving system quality and robustness, and reducing management overhead. The toolkit we developed generates a complete running system including APIs and user interfaces from the system definition written in our DSL. This was a substantial project which is now being prepared for production use.
Towards measurement for source code quality
January 1, 2015 – December 1, 2015
The goal of this project was to develop a statistical model of source code. A model that goes beyond syntax, and by learning from a large collection of “well written” code, attempt to model code semantics at a rudimentary level. I approached this task by building a model that extracts patterns present in the code it is trained on, treating patterns in code as a proxy for semantics. Once trained, the model would give higher scores to unseen code that has similar patterns. This research was based on finding a good decomposition of the code’s Abstract Syntax Tree (AST) - a decomposition that does not loose too much information about the long distance relationships present in code. Our model encodes only the structural information of the AST, and not the surface tokens of the program code, and once trained gives lower scores to code similar to the training set. We tested this model across two different programming languages (python and Java), on 7 different projects. Used python source code belongs to two different domains (scientific computing, web) and, used Java source code belongs to 3 different domains (database, build tools, search engine). In every case, given previously unseen code, our model was able to identify both the project and the domain with 100% accuracy.
Step Climbing Robot
September 1, 2013 – Present
We built this robot for a competition at SLIIT. It is capable of climbing steps that are 5 inch tall and can follow a line to get to that object that it is suppose to pick. We build a hand into this robot to pick up the object from a specified destination and drop it at a specified destination. There were few junctions on the way as well.
GPRS Activated Door Answering System
September 1, 2013 – Present
It's basically a door answering system with the capability to call the owner of the house if they are not at home. This system can be activated to work with both the GPRS technology and through the landline. This system will be a very important thing to the society with the busy schedules, because lets say if somebody come to your house to give an urgent message and you are not at home, and that person doesn't have your mobile phone number. What happens then is it will be too late when you hear the message. But with this system in place that will not happen, because if you are not at home system will automatically give you a call connecting that person with you through the in built speaker and the microphone, that way you will get the message then and there. So we think it's an essential system for the society nowadays.
Train Platform Direction System using an FPGA
March 1, 2013 – Present
The system will automatically detect whether a train is coming using weight sensors and it will automatically close the gates in the railway crossing that way we can avoid accidents in the railway crossings. This automated gate system is important because if we assign a person to do that job it'll be a waste of human labor because he won't have much work most of the time in the day. And by using this system we can avoid human errors as well. Another importance in this system is that when it comes to a train station it will show the train driver which platforms are free for him to go to. We can reduce human errors in that specific task as well by automating that. We were trying to automate the already available systems to increase the accuracy of those tasks, that's the main reason we used a FPGA to do this task as well. Because we can do high speed computing using an FPGA so the system will be very accurate.
Cultural Fit Analysis
The candidate's project history shows a strong inclination towards research and development, particularly in AI/ML and complex system design. The transition from engineering roles to a product management rotation at ASAPP, and prior experience as a Director of Technology, indicates a versatile individual who can adapt to different roles and responsibilities. The breadth of projects, from academic research to practical applications like the GPRS door system and robotics, suggests a proactive and innovative mindset. However, the target role of 'Data Analyst' might be a step down from their current 'AI Research Scientist / Research Product Manager' role, potentially indicating a mismatch in career trajectory or a desire for a different focus. The projects are highly technical and research-oriented, which aligns with a problem-solving culture, but the specific 'Data Analyst' fit needs further exploration.
Soft Skills & Operational Fit
The candidate's project descriptions suggest an ability to identify problems and propose innovative solutions. Experience in cross-functional collaboration (ASAPP, WSO2) indicates teamwork and communication skills. The diverse project portfolio, including robotics and embedded systems, suggests adaptability and a broad technical curiosity. However, without specific psychometric test results, a detailed assessment of work attitude, stress handling, and team collaboration is not possible.