Hyperion.ai allows users to track any metric that they like over time. Nowadays, there are a ton of new health fads, diets, workout routines, etc. Often times these fads are not supported by any significant evidence. Additionally, they often rely on solely on the experience of the individual who happens to be sharing the information. However, people are all different. Something that works for one person does not necessarily work for someone else. By using Hyperion.ai, users can quickly narrow down on what works and what doesn’t. Hyperion.ai will graph the users' metrics onto custom charts that they can define. Additionally, because it is sometimes difficult to determine correlation between several metrics, Hyperion.AI uses a deep neural network to help find these correlations and present them to the user.

You can check out a demo and writeup of the deep network here: http://alx.lu/ml

And the marketing page can be found here (note: demo vid on site is outdated): http://www.hyperion.ai

I created Hyperion.AI using React, React Native, Redux, and Redux – saga. The backend is written using node JS, Redis, and PostgreSQL. It is deployed to Heroku along with a custom clock process that will send unprocessed data to the deep neural network that I deployed to AWS.

Hyperion.ai was initially part of a startup that I had co-founded full time. We were initially planning to ship a supplement that would use our deep network to continuously tweak the user's supplement formulation based on their metrics.

You can read the original write up for the startup here: http://alx.lu/work/hyperionai-old

While it was a great learning experience, we realized that it would be incredibly difficult to prove the efficacy of the physical product before launch. We did not have the resources to run any sort of clinical trials on the product and therefore we felt uncomfortable releasing the supplement to the public.

I ended up shipping the app as a standalone product. This went well for a few months and the product experienced some solid growth. At that point, there was still no solid business plan. Since I was working on the startup by myself at this point, most of my time was spent trying to stabilize the service to accomodate the influx of new users. Ultimately, the service had to be shut down because it became untenable to justify the cost of maintaining the service without a concrete plan for generating revenue.

I believe there is still potential in this concept. It's something I will definitely consider revisiting in the future.

Alex Lu 2016