Who ever said a Virtual Hackathon wouldn’t be possible?
Datalex held its annual hackathon during the week of February 1, 2021, remotely, and with great results.
The primary objective in organising organising the event was to provide the Datalex team members with the opportunity to work on new and innovative approaches aligned with the company’s product vision and working environment. In preparation for the Hackathon, the challenge was set to explore creative solutions and innovations focused on:
• Datalex Technology
• Customer & Market Facing Product Capabilities
• Employee & Teamwork Environment
Preparation for the Hackathon included soliciting a series of proposed themes and ideas which were refined into the main projects for the hackathon. Project leaders were then chosen for each of the Hackathon projects.
The Hackathon week was launched on Monday February 1st in a fun and engaging Opening Ceremony during which project leaders presented a pitch and description of their Hackathon project. Hackathon participants then chose a project to work on. Teams broke off to organize and plan activities for the week to develop their ideas and innovations. Final presentations were made to the judges on Friday, February 5th and the winner was announced the following Thursday.
With the current work from home requirements, the teams were faced with the challenge of not being able to brainstorm, whiteboard and work as a team as would in a ‘traditional’ hackathon. However, the experience of the past year working via remote tools ensured that they were all able to overcome this obstacle and collaborate very effectively.
An excellent slate of projects was developed and presented during the hackathon. Several of the teams, including the winning team, focused on Machine Learning and Artificial Intelligence concepts (ML/ AI) concepts and solutions. These technologies are recognised as critical for success in the airline industry. Datalex is committed to implementing ML/AI capabilities across the Digital Commerce Platform.
Teams were judged on these four criteria:
1. Business Value
2. Level of Innovation
3. Knowledge Expansion / Learning
4. Completeness
After a careful review of all projects, the judging panel* selected the winning project: Machine Learning – Offering Relevant Ancillaries Using Machine Learning and AWS
The winning team was led by Philippe Tap, and team members included Alberto Galan, Brian Staines, Desmond Lewis, Jacky Chan, Paul Lester, Rod Osborne and Ellen Tracy.
It is clear from this year’s projects and presentations that machine learning and different forms of artificial intelligence are in. Three hackathon projects focused on unique opportunities where Datalex can leverage ML/AI technologies, as a requirement for driving revenue and enhancing customer engagement for airlines.
The winning team proposed using Machine Learning and Artificial Intelligence to recommend personalised offers throughout a travel consumer's engagement with the airline and at various touchpoints in the engagement cycle pre, during and, post trip. The team also focused on technology to actively reach out to travellers at various stages in their journey to present relevant offers to enhance and round out the traveller’s trip experience.
The winning team focused on how machine learning could be used to overcome the constraints of traditional static rule-based offer management solutions by extracting patterns from training data fed to the machine learning toolset. A machine learning based offer engine can scale to deal with hundreds of predictive features across thousands of products. The team identified that leveraging AWS technology like Amazon SageMaker, AWS Glue and Amazon Athena could significantly accelerate the development of a machine learning based offer generation solution.
After multiple rounds of review, we think this team has come up with a very interesting set of capabilities and potentially some “secret sauce”. So, we decided to keep some of this close to our chest while we investigate and refine the concepts and capabilities to turn them into a solution set that can drive value to Datalex customers. Stay tuned for more details.
The team found that having a diverse group of disciplines represented across its members helped refine the concepts and kept them focused on a Minimum Viable Product (MVP) type of approach for the initial phase of a machine learning project. Members included developers formally trained in machine learning, developers with extensive airline experience, network administrators, solution architects, and a high-level business and finance manager with deep financial modelling experience. Everyone contributed to the development of the concept and the presentation.
The idea for the hackathon project initiated with Philippe after his recent 12 month-postgrad program in data analytics and his work with a machine learning working group focused on preparing training for full-stack developers. Also, Philippe had an experience with an airline where they sent him an offer for a car rental even though he had already purchased a car rental from the airline. He thought they missed an opportunity to offer him a more relevant product that he might want and purchase.
The initial idea matured over a few weeks before the hackathon and was presented to the team on the first day. Team members identified weaknesses in the proposed solution, and this helped refine the problem statement and evolve the solution to its final and more robust form.
To complete their project, the team took a divide and conquer approach to research and building out the different aspects of the project. Analysis and work were allocated across the team as follows:
Philippe: Concept innovator and shepherd. Machine learning expert. Overall project management and capability definition.
Jacky: Researched and defined how to fit the solution into Datalex Digital Commerce Platform.
Rod & Ellen: Focused on financial projections for the airline and Datalex including cost savings, revenue enhancement.
Brian: Researched how airlines are attempting to target offers and similar capabilities that are implemented by airlines and competitors.
Paul: Reviewed the proposed solution and determined how it fits in the current technology stack. Investigated additional data sources to refine the learning model.
Alberto: Developed detailed annualised cost analysis of the solution including AWS services.
All aspects of the research were reviewed as a team. The final piece of the puzzle was the presentation. The team agreed to remove unnecessary details to keep the presentation clear, concise, and punchy. And it seems to have paid off.
In choosing this project as the winner, the judging panel commented that the team’s concept and presentation were thorough, innovative, focused, and viable. Datalex has committed to allocating resources to fully investigate the capabilities defined in the project and to potentially implement this capability set in the Digital Commerce Platform.
* Datalex Hackathon 2021 Judging Panel: Bobby Healy, Sean Corkery, Conor O’Sullivan, Andrei Grigoriev and Ryan Estes.
|
Datalex Hackathon |