Our team developed a smart travel booking app powered by ML & Big Data technologies, and helped the Client cut costs through infrastructure optimization and migration to DigitalOcean.
- BUSINESS NEEDS: The Client is a promising startup in Travel & Tourism and was preparing their platform for a beta presentation to get new investments. They engaged Tech Web Support specialists to fine-tune the product & feature it with additional functionality.
- RESULT: The team performed UI redesign making the platform more user-friendly. Our ML engineers implemented an ML module with personalized recommendations & pricing trends forecasting. Moreover, we optimized the Data Platform and infrastructure thus reducing product maintenance costs by 30%.
2020 WAS A TOUGH YEAR FOR THE WHOLE TRAVEL INDUSTRY. WE PUT ALL THE EFFORTS TO HELP OUR CLIENT TO OVERCOME THE CHALLENGES OF THE NEW NORMAL.
Having analyzed the solution, we focused on improving its UX (both from UI & recommendations perspective) to increase user engagement, architectured Big Data injection, and implemented ML module that extended functionality and simplified trip planning for end-users.
2020 turned out to be a tough year for the whole travel industry, especially for startups that were looking for the next rounds of investments. So, when our Client faced a similar situation, we also helped them to perform DevOps transformation to lower expenses and make the startup leaner.
- Delivered an attractive and engaging UI according to the best UX practices
- Designed and developed the Big Data platform with data pipelines optimized for working with a Machine Learning module
- Brainstormed and developed a smart and simple Machine Learning model for generating travel recommendations
- Migrated the product infrastructure to the DigitalOcean to lower online booking solution infrastructure costs while staying within the same performance KPIs
BIG DATA AND ML:
The Big Data part was a pleasure cruise for the team – thanks to our expertise in Big Data & an in-house framework we elaborated the robust and easy-to-maintain Data Platform fully optimized for fast Machine Learning model execution.
However, the ML solution turned into a challenge since we lacked data for predictive algorithm training (only 2 months’ statistics were available while we needed 10 times more). We perused numerous public researches and suggested an alternative algorithm for price prediction that could work with the limited data and gave better results.
- Big Data platform
The Big Data team defined the future architecture of the Data Platform and developed the solution keeping the focus on the Client’s objectives, limited budget, and short timeframe.
- Machine Learning Module
Our Machine Learning developers designed and developed unique price trends prediction Machine Learning algorithm and optimized its training process decreasing hypothesis testing from 2-3h to 5-10 minutes.
Another major goal was to help our Client cut infrastructure and product support & maintenance costs for the travel booking system. We made a set of improvements and migrated the infrastructure to the new budget-friendly cloud solution.
Our DevOps team also rewrote the existing charts in Helm 2 (Kubernetes apps management system) to the Helm 3 ensuring multi-cloud support. This allowed us to migrate the travel management platform from one cloud provider to another with the lowest downtime and safeguard minimum expenses.
- Created a scalable infrastructure supporting CI/CD processes with its own monitoring and logging
- Implemented Infrastructure as Code (IaC) through Terraform-based infrastructure modules
- Successfully migrated the global booking solution from Google Cloud Platform to DigitalOcean for more cost-efficiency
- Applied GitOps practice (Terraform + Helm3 + GitLab CI) enabling small teams to faster deliver the product increments to test & production servers