Good Practices for AI-Assisted Development: Lessons from a Live Protein Calculator Demo
Alternatively titled: Adventures live coding in a conference setting... with AI!
Hello fellow datanistas!
Have you ever wondered how AI can transform your coding workflow?
In this blog post, I share insights from a live coding demonstration I gave at BioIT World 2025, where I built a protein mass spectrometry calculator tool.
During the presentation, I showcased how standardization in project structures can actually unleash creativity rather than constrain it.
I started with a design-first approach, using AI to help draft a comprehensive design document before diving into coding. This method not only accelerated development but also ensured that I maintained oversight and applied software development principles throughout the process. There were some humorous moments from the demo, like when I had to switch to 'cooking show mode' and pull out a backup repo already created (with AI assistance too!) due to a package dependency issue.
The key lessons learned include the importance of context, interactive communication, and documentation in AI-assisted development. These principles can help data scientists develop more maintainable and effective software solutions.
In the end, using AI thoughtfully can dramatically accelerate development while allowing us to apply our domain expertise and judgment. It's all about maintaining human agency and using AI as a powerful assistant.
I invite you to read the full blog post here. If you find it insightful, please share it with others who might benefit from these practices!*
Happy Coding!
Eric
P.S. One of my friends, Stephen Goldfless, is hiring for a Director of Data Science at Lila Sciences. If you’re interested, check out the job post here!