Valui - Testing a MVP with a predictive model
By middle Q3/2017 a friend of mine contacted me about an idea he would like to validate. We met and he explained his idea for an online service on the real estate market. It was an ambitious and very multidisciplinary project.
Initially it was a data science / machine learning job, deployed to an endpoint on the cloud. I knew I didn't had all the skills by that time, neither the time to freelance that all by myself, but I won't let a friend down. I turned to help him, but at a new level. I'd outlined the professionals and expertise needed to get this as better as it could be; this gave us a budget idea.
I was committed to the cloud deploy and also charged a fee over the consulting, hunting and managing services. With that approved, I looked for the professionals - a senior backend and a data scientist - and we started. The senior backed dev was responsible to get the data from multiple sources and consolidate them into a single ordered source, so the data scientist could do his work.
The data scientist created the models and validate them along with the client. When we get a good enough model it was my turn, I deployed that model into a lambda function on AWS, and exposed via api gateway. The website project he was already conducing with an agency could then use the service we created.
But it turned out that this website project was taking too long to get ready, and it was already over budget... My friend asked me if I could do something about too. I followed the same process... I outlined and hired one UX specialist, one frontend developer, one backend developer, and with a lot less cost, we delivered an amazing website, up and running.
Finally, he asked if I knew someone who could help with campaigns on google and facebook. Of course I put him in contact with a another friend, and he could finally run his experiment and gather metrics, which enabled him to go further with a business plan and investor seeking.
This is a different story, where I used my knowledge and network to outline, define budgets, hunt, hire and manage a complex project. There were 8 people involved in total, and at least the experience and metrics gathering for validating the idea were successful.
Would you like that too? Hire me!