A Day in the Life of a Machine Learning Engineer at Agoda
Machine learning engineers are at the forefront of technological innovation, blending software engineering and data science elements to develop systems that can learn and adapt without continuous human oversight. At Agoda, these engineers leverage advanced analytics to enhance user experience and operational efficiency. In this article, Borvornsak Laoratanapong, Senior Machine Learning Engineer at Agoda, shares insights into his role and how he leverages advanced machine learning techniques to drive strategic advancements at Agoda.
From Intern to Engineer: My Career Path in Machine Learning
My journey at Agoda began over six years ago as a summer intern. Initially, my task involved developing a tool that helped engineers quickly access Spark application logs. This experience was pivotal, setting the stage for my transition into machine learning engineering. After completing my internship and graduating, I rejoined Agoda as a machine learning engineer on the same team where I had started. Today, I help implement, test, deploy, and monitor automated bidder systems for managing billions of ads. My work involves extensive A/B testing across various online advertising channels, including search engines like Google and Bing, as well as meta-search engines like Tripadvisor and Trivago.
My Role as a Machine Learning Engineer
In the data or machine learning area, you might hear about three collaborative roles: data engineer, machine learning engineer, and data scientist. Machine learning engineers are the ones in the middle. They need to work with data engineers and scientists to maintain the data pipeline and productionize machine learning systems to solve business problems.
Maintaining and managing billions of ads daily requires significant human effort. This is where machine learning-based automation systems come into play to help us efficiently produce profitable bids for each specific ad for hotels in Agoda against all other OTA competitors.
Embracing Data-Driven Innovation
In my daily tasks, I apply various machine learning techniques to optimize our advertising bids and efficiently manage a high volume of ads. For instance, through reinforcement learning, I tackle the unpredictable dynamics of our partners’ advertising platforms, allowing us to adjust our strategies in real-time.
Handling and analyzing large data sets is complex, requiring clear and specific requirements to guide our search for insights effectively. For instance, we utilize Apache Spark in our big data ecosystem for its ability to rapidly perform large-scale data processing and analytics, thanks to its implicit data parallelism and adaptive query execution. These tools enhance our ability to handle data efficiently and support our machine learning algorithms across various programming languages.
Enhancing Collaboration Across Teams
Effective collaboration is essential at Agoda, particularly between technical and non-technical teams. Our team employs various collaboration strategies, such as system design reviews, task breakdowns, and code reviews with data platform and data backend teams. These interactions are vital for aligning creative ideas and gathering requirements for new marketing strategies.
As a data-driven company, our collaboration extends beyond the technical teams. We engage with marketing teams to ensure our machine learning objectives align with broader business goals. When we want to experiment with a new marketing campaign strategy, for instance, we gather requirements from the marketing teams, discussing the business ideas in depth to assess feasibility and resource requirements. Our goal is always to leverage our engineering capabilities to find the most effective solutions, ensuring we meet the strategic needs of the business efficiently and effectively.
Success Story: Optimizing Ad Structure for Enhanced Performance
We significantly reorganized our ad structure to reduce the number of ads requiring maintenance and enhance the performance of our machine-learning models. This strategic update allowed our models to learn more effectively from the performance data of the new ad structure.
Beyond improving algorithms, we emphasize data cleansing, manipulation, and processing. Our approach demonstrates that quality input is critical— better input consistently leads to better output, even when using the same machine-learning techniques.
Maintaining Work-Life Balance when dealing with complex projects
When managing challenging projects, I remember that I am part of a team. Taking ownership doesn’t mean doing everything independently; it involves overseeing the planning, execution, and timely delivery of quality work. I can delegate tasks to my trustworthy team, who are always willing to help.
Staying Ahead: Innovation and Research at Agoda
I make it a priority to stay current with technological advancements by participating in internal tech talks, interacting with colleagues about new tech developments, and exploring real-world machine learning trends. Recently, we hosted an in-house generative AI hackathon, diving deep into this emerging technology to uncover its potential applications at Agoda. These initiatives provide insights into practical applications of cutting-edge technologies and stimulate creativity within our teams, leading to innovative solutions that keep us competitive.
Future of Machine Learning at Agoda
Machine learning is evolving consistently at Agoda, and there are several indications of how it might develop in the coming years. Generative AI has been the most popular topic in the last few years. There has been substantial interest and advancement in powerful ML models like ChatGPT, Stable Diffusion, and others, which provide the potential for many interesting use cases.
Beyond new trends, we keep improving ML in our project successes and expansion. Projects such as the Ranking team’s use of a Recurrent Neural Network to recommend hotels based on user clicks and bookings and the Anomaly Detection project, which employs time-series ML to monitor various important metrics, show how ML brings value to Agoda. These initiatives demonstrate the ongoing value of ML at Agoda and are expected to advance with new algorithms and sophisticated models.
Advice for Aspiring Machine Learning Engineers
If you want to pursue a career in machine learning at Agoda, my first suggestion is to be a good software engineer. Building a solid fundamental skill in software development will help you quickly get into a machine learning engineer role when dealing with data-related problems. Communication skills are also crucial. We have to work with a wide range of people, including non-technical ones from the business side. Ensuring flawless communication and collaboration among all stakeholders is a must.
I want to share one last thing with you: ‘There is no dumb question; you just don’t know it yet. Don’t be afraid to ask questions.’ I learned this when I first started here, and it opened up many opportunities for me. I hope it also helps you unlock that door full of opportunities in your areas of interest. :)
