CIO100 2018 #31-100: Andrew Bates, Harmoney
“Harmoney, Australasia’s leading marketplace lender, is at its core a data company, we are continually looking for new technologies to support a data-driven, customer-focused user experience,” says Andrew Bates, CTO at Harmoney.
“Two key parts of our current technology programme are good examples of our approach to delivering real business value.”
The first part, he says, is a new adaptive technology platform designed to receive, process and analyse rich customer data in real time. The objective is to reduce the amount of data we ask the customer for, especially where we already know the answer, says Bates.
This means we have made optimal use of the available data, analysed and processed it in real time driving insight and leading to a much more dynamic customer experience, he adds.
The second part is a machine-learning platform for data science.
We are combining a data science platform and a separate data management package to bring together software engineering and data science teams on to a single platform, he explains.
“This is an exciting outcome for a business like ours in reducing the cycle time of a new development from three months to one or two weeks,” he says.
“Rather than an inefficient ‘Insight to Action’ process (which can be manual and slow) from data scientist to business analyst to engineer to tester, with multiple hand-off points and interpretations of data, we can cycle through the process a lot quicker and get to the outcome we want four to six times faster, all whilst reducing delivery risk.
“We can now feed data into a real-time data pipeline to a tool that runs a huge number of machine-learning algorithms to discover relationships between data sets; execute multiple permutations; and build a machine-learnt model that tells us more about our customers and how to better serve their needs. Agility is essential, and the ability to operate within unknowns, with the skills to adjust, adapt and innovate moment by moment, is the culture of these times.”
Bates says supporting this innovation is a move towards a microservice-style architecture, based on the Netflix open source stack, and incorporating technologies such as Apache Kafka to deliver real-time data pipelines, with two off-the-shelf packages covering machine learning and data management supporting both traditional batch and real-time processing of data.
“This approach supports a unified deployment and management model with new data pipelines, data transformations and integrations deployed as a set of microservice components,” says Bates.
This innovation enables Harmoney to move from batch-driven to real-time data pipelines, enabling trained machine learning models to be executed with online customers and ensuring Harmoney uses a common set of tools to deliver a data-driven company.
A project they recently kicked off was looking at how conversational chatbots can help the inbound customer service experience.
Bates explains the old model was a human based contact centre and chat experience ending at 8pm. If a customer calls after that, they are directed to email or voicemail for follow-up the next day.
They are introducing a chatbot that can continue to support the customer interaction 24 hours a day, so the customer can choose how they wish to interact with us.
“We can see keywords and answer common questions directly or provide links to FAQs and relevant knowledge base articles.”
The chatbot will enable them to reduce calls into the contact centre, improving operational efficiency and freeing resources to focus on outbound calls.
Bates is part of the senior management team, which has a weekly Monday morning session.
“This is a good opportunity to understand where my team and I can add value and drive automation, so I am always looking for signs of manual processes or issues we can help resolve,” he says.
“We have a company-wide stand-up on Monday mornings as well, where we cover what is being put into production. We use this time to also celebrate success among the team.
“I am the product owner for an R&D team of eight here, Harmoney Labs, which is currently building out our ‘adaptive platform’.
“This platform will make better use of existing data we already collect; introduce new datasets and technologies; and provide interesting opportunities to add additional value to what has traditionally been a transactional conversation with our customers.
“This approach is also moving us to a DevOps-style culture, with high degrees of automation across infrastructure setup (Infrastructure as Code) and build pipelines, and continuous integration and automated deployment/release management.”
He says his overall role as a leader has changed in the past three months and he could now get back into focusing on strategic projects, design and their technology roadmap.
Previously, he had around 20 people reporting to him, which made his role “very managerial and tactical”.
“Now I am free to work on our technology roadmap and research and development, which reflects a maturing of Harmoney as a business,” he says.
“Over the past six months we have also grown to over 30 people in the engineering team. We were running two scrum teams and then scaled it to five cross-functional teams that are closer and more autonomous in how they operate.”
The biggest lesson in his career, he says, is the need for ICT leaders to continually improve their communication skills to lead/guide teams of talented people.
“This learning doesn’t stop and can always be improved,” he says. “In addition, while juggling multiple priorities and responsibilities, technical leaders need to take the time to communicate effectively and ensure the team is also on the journey, we explain the ‘why’ and engage them to help us find better solutions.
“As an organisation changes, grows and matures, this skill often needs to adapt to what the business/team requires at that moment in time.”