Open source project consisting of handy machine learning tools built like an extension of scikit-learn. The components are built according to scikit's object API and automate things I kept repeating for feature engineering, cross-validation and ensembling.

Only uses scikit dependencies, allowing me to focus on working more with numpy. I mostly use this for Kaggle but they are generic machine learning components that are quite useful.

Tech:    [Python, scikit-learn, numpy]

Open source ball & beam simulation as OpenAI reinforcement learning environments.

The main aim of the project is to allow comparison between reinforcement learning and conventional approaches from control theory, such as PID controllers.

Tech:    [Python, numpy, matplotlib]

At work we want to name our projects after pokémon, but our legal department did not like the idea. To solve this I built a name generator that generates truly unique pokémon-esque names.

It is a Char-RNN based on Andrej Karpathy's famous design. Built in Keras, implemented as a generic char-level LSTM text generator that can generate any text it is trained on.

Tech:    [Python, Keras, Tensorflow, Pandas, numpy]

Kaggle is a platform where people in data science can compete for the best solution. Kaggle also has a nice community where it is possible to learn from the best!

Santandar Customer Transaction Prediction – Top 9%

Home Credit Default Risk – Top 19%

Tech:    [Python, numpy, Pandas, scikit-learn, LightGBM, Keras, Tensorflow]

Stack Exchange is a collection of sites where people can ask and answer domain specific questions. Stack Overflow - every programmers saviour - being their flagship.

I am active on the Stack Exchange devoted to data science. There I try to answer other's questions to the best of my abilities. I find that you learn a lot yourself when you try to help others.

Rank: Top 1%, this year

Elm is a small purely functional Haskell-like language/framework for web frontend. After finishing a work project I took what I had learned about how to effectively build a SPA in Elm and made an open source template.

The template has been used by both companies and people who had just gotten into Elm and wanted a rapid start with a good scalable foundation to build upon.

Tech:    [Elm, Javascript, HTML]

Inspired by Getting Over It, Stickboi is a frustrating game where you control a stick named Stickboi. The game has been described as "the closest you can come to experience how it feels to be a stick".

In development, working together with a friend. It is built just for fun and to learn a bit of game development.

Tech:    [C#, Unity, Python]

JIRA is a system that, amoung other things, runs alot of companies internal ticket systems. In large companies it can be hard for user to know how to properly assign their tickets. To solve this we built a ticket auto assigner using machine learning.

The tool reads the information that the user has put in the ticket and uses NLP methods to automatically assign categorices such as resolution group, urgency and impact.

Tech:    [Python, NLP, scikit-learn]

A webshop had issues with a large number of human errors in their database that led to customer complaints. The solution was to continuously train a machine learning model on items with errors and then scan the database to find unknown errors.

The core is an ensemble of both supervised and unsupervised machine learning methods.

Tech:    [Python, NLP, scikit-learn, Tensorflow, C#, .NET]

Consulting for a large scandinavian bank who bought a new factoring system. The project was to integrate it into existing application eco system and giving it a frontend.

Tech:    [Scala, Ruby on Rails]

Web application for a large Scandinavian bank where loan and leasing contracts can be finalized with a buyout. Filled with fun financial calculations and models.

Functional programming top to bottom with a Scala API backend and Elm SPA frontend. I was responsible for building a SPA architechture which I later based an open source project on.

Tech:    [Elm, Scala, Jenkins, Docker]

Seavus is a large tech consultancy with offices around the globe. Having various different areas of expertice such as finance, gaming and recently AI.

I work as part of Seavus' DoAI team. We build custom machine learning solutions for customers in all kinds of industries.

Tellox is a small consultancy that holds a lot of knowledge in the domain of banking software. They employ many of the architects of the old COBOL banking systems that still are in use today.

I worked with building web based financial applications, often replacing old applications running on COBOL.

B&R was previously the worlds largest privately owned automation company before it was bought by ABB.

Started as a trainee in Austria and then went home to be an Application Engineer consulting in Stockholm. Majority of my time at the company was spent doing robotics and motions control applications.