Forem: Florian Treml

We are using Javascript for our SaaS-Backend

Florian Treml — Fri, 29 Jan 2021 10:29:06 +0000

For our SaaS-product Botium Box - a full functional and non-functional testing infrastructure for chatbot - we decided to go for Node.js in the backend.

There are lots of controversial arguments about programming language in general, and using a non-type-safe, dynamic scripting language like Javascript for server-side development. In our company we have a strict no-bullshitting-policy, so we didn't go very deep into the more philosophical aspect of selecting a programming language.

After all, it came down to a decision Python vs Javascript. While Python is by far the most mature language for machine learning projects, the one single argument why we decided for Node.js was that in our small developer team (4 people) everyone is a full-stack-developer, and it will ultimately safe time to develop client and server in the same programming language.

P.S. For some machine-learning-centric components of our SaaS-backend we still decided to go for some micro-services developer in Python.

No, You Are No A.I. Company When Building a Chatbot

Florian Treml — Fri, 22 Jan 2021 15:09:36 +0000

According to the report The State of AI 2019: Divergence there is a lot of Bullshit Bingo around the thing called Artificial Intelligence (AI). From page 101 of the report:

“In approximately 60% of the cases (…) there was no evidence of AI material to a company’s value proposition.”

This means on the other hand that 40% of AI companies are not really making value of AI technologies. Product taglines are full of buzzwords to make it appear interesting. For example this one:

“With the help of Artificial Intelligence, Machine Learning and the Blockchain our chatbot organizes a meetup with all your friends”

What this product actually does:

It connects to various calendar applications
It finds a free spot for you and your friends
It blocks this spot in the calendar applications

While such a product might make sense, it has nearly nothing to do with AI (and I don’t see anything Blockchain-related here, but that’s another story). The user interface is a chatbot application based on natural language: “hey xy-bot, organize a dinner with my closest friends next week”. Surely, there is intent and entity resolution involved, but as long as this company doesn’t work on it’s own NLU engine, which it most likely doesn’t, I don’t call this company an AI company.
Using an NLU engine doesn’t make you an AI company.

How We Are Saving Costs By Dumping AWS Cloud

Florian Treml — Thu, 21 Jan 2021 16:59:22 +0000

Amazon is using the same trick with AWS pricing as cellphone providers: they rely on bad math capabilities of their clients. The costs for one single item, in the case of AWS this is one hour of computing, is incredibly low, parts of a cent - but most people out there are not able to do simple maths like multiplication (by 720, the number of hours in a month) or summation (with the costs of the other services you have to implicitly use).
When hosting a simple one-node Kubernetes cluster the costs are easily at a three-figure number per month - the fees for EKS, the fee for the EC2 node, the fee for the inbound load balancer, the fee for storage, ...

Let's do a simple calculation for a Kubernetes cluster:

$58,40 for one EC2 instance (r5a.large, reserved)
$74,40 for EKS (Elastic Kubernetes Service)
 $9,00 for EBS (Elastic Block Storage)
$20,00 for ELB (Elastic Load Balancer)
... and some minor costs for data transfer and VPC

So even the most simple Kubernetes clusters has costs of around $150 per month - for computing power, open source software and inbound traffic.

We decided to switch to another baremetal provider pretty quick - the costs are $11 for a root server that outperforms the EC2 instance for $60 easily.

Amazon is selling shovels in the gold rush: overpriced computing power for unprepared startups.