Hackathon 2025 Recap: Innovation, Execution, and the Spirit of Competition
Last year, we launched our first ever engineering team Hackathon, which is where we dedicate a whole week for our developers to step away from their daily tasks, experiment with new technologies, and tackle business challenges. The innovative ideas developed during last year’s Hackathon sparked creativity and led to real solutions that are now part of our systems, so given its success, we brought it back this year. This focused time allows the team to build prototypes, foster collaboration, encourage bold thinking, and strengthen our engineering culture.
The Game Plan
For Hackathon 2025, which occurred in January, we proposed three main challenge areas to guide everyone’s thinking:
- Automating Manual Processes: Cutting down on repetitive tasks, reducing errors, and boosting efficiency using automation, AI, or machine learning.
- Optimizing Technical Architecture: Looking for ways to improve system performance, lower costs, and explore modern frameworks.
- Understanding Customer Behavior: Analyzing user activity and predicting which free-trial customers might convert to paying customers.
Teams could pick any problem within these themes—or something closely related—and then spend a few days planning before diving into the one-week sprint to build working prototypes.
Noteworthy Projects
Even though many projects were developed, a few stood out for their impact and originality:
- Automated Batch Testing: One group tackled improving our test-driven development framework for the Batch Upload feature. They built a framework that automatically checks batch output files for errors and integrates with our existing Postman and CI/CD pipelines. By storing files in a secure bucket and handling batch tasks through dedicated APIs, the new setup cuts down on manual testing and makes the whole process more reliable and scalable.
- Exploring Reactive Programming: Another team focused on improving the performance of our APIs by experimenting with reactive programming. They swapped out blocking HTTP calls for a non-blocking, asynchronous approach. Initial results were mixed due to some dependencies that still used blocking calls, but a second round of tests with a fully non-blocking setup showed clear gains in throughput and latency. While not ready for immediate development and deployment, this work lays the groundwork for future architectural improvements.
- Data-Driven Conversion Predictions: A different project dove into user behavior to predict which trial users might become paying customers. By analyzing historical data—like how often users interact with certain features, how they engage with our APIs, and whether they keep using our tools after their trial ends—this team built a machine-learning model with over 80% accuracy. They also added a dashboard to visualize insights, making it easier for sales to prioritize leads.
- Customer Engagement Scoring: One more initiative took a broader look at customer engagement. This team pulled data from multiple sources, scored users based on their activity quality and breadth of feature usage, and grouped them into high, medium, or low-interest categories. They even built a way to tweak the weighting of each metric, giving the sales team flexibility to adapt the model as business needs evolve. It’s still in the prototype stage but offers a promising framework for deeper customer analysis.
Recognizing Excellence
While every solution added value, we highlighted two special awards:
Excellence in Execution
Given to the automated batch testing project for delivering a production-ready solution within a week. It’s already been deployed in our pipeline, saving time and reducing errors.
Innovation Catalyst
Awarded to the conversion prediction team for their inventive use of machine learning to help us understand which trial users are most likely to convert, complete with a user-friendly interface for the sales team.
What’s Next?
Automated Testing → Already in Production
The testing framework has been rolled out and is now part of our daily workflow. It improves accuracy in batch processing and saves valuable developer time.
Merging Data Insights for a Better Conversion Model
We see potential in combining different customer analysis efforts to create a more powerful predictive tool. By merging the conversion model with extra insights from the engagement scoring project, we aim to fine-tune accuracy and give sales an even sharper way to identify promising leads.
Future Considerations for Reactive Programming
Although the reactive programming project isn’t live yet, the proof-of-concept results were encouraging. We’ll consider these findings as we evolve our tech stack and look for ways to boost performance in the future.
A Tradition That Keeps Growing
Each year, Trestle’s Hackathon proves that when you give engineers the freedom to explore, the results can be encouraging. It’s an opportunity to challenge the status quo, break routines, and spark innovation that drives our business forward.
With one project already live, another poised to refine our sales strategy, and a pipeline of future enhancements taking shape, it has been an extremely productive week. We’re excited to see how these ideas evolve and even more excited for the breakthroughs that next year’s Hackathon will bring.
