Path to ML - The Genesis and the Journey

Originally, our journey at Sensure began as a humble hackathon initiative. We were driven by the audacious goal of disrupting the insurance arena by leveraging Machine Learning (ML) and Artificial Intelligence (AI). Envisioning an open-insurance policy with the capacity to self-assess risk and a public risk profile, we engineered the skeleton of our grand idea. Our creation? A motley assortment of code repositories, each component meticulously crafted using the best stack for the job. A true ala carte of code.

The Python Prelude

An impressive number of services in the Sensure stack owe their genesis to Python. After all, Python is the darling of the data engineering world. If you're still stuck in the past, remembering Python as the framework that fueled the freeware glut in the Windows-PC era, you're not alone.

The Python Farewell

As time passed, my once bustling band of team members, who'd been with me since the hackathon, dwindled. As I traversed through the professional landscape, hopping from one job to another, I inevitably ended up working with .NET-centric teams. The logical choice was to morph our architecture into a stack that I could comfortably transition to during my off-time, aligning it with my daily grind. The Python codebase and Jupyter notebooks languished, forgotten, in their respective repositories. Until...

The .NET Greeting

Enter ML.NET. I've been observing Microsoft's endeavors in the AI and ML space for quite some time, yet always held back from migrating my existing machine learning code to .NET. Why? Simply put, .NET hasn't quite reached the gold standard of being a first-class citizen within the data engineering community. With its relatively quieter activity, it's more prudent to stick with the bustling Python community when dealing with data work. I won't bore you with the details. If you're here, I assume you share my affinity for .NET.

Going forward, I'll be writing about my plans to integrate ML.NET with some of the existing ML Models in the Sensure stack, starting with the required transformations within a .NET Core project. Although these posts might not follow a strict sequence, they'll address the challenges that could arise in a generic manner.

If you've been toying with SQL Server and EFCore in your .NET Core projects and are contemplating migrating to PostgreSQL, head on over to my first post in the series, where I guide you through setting up a time series database. Taking a Leap into Time Travel

Let's start this journey together.