Martin Singh-Blom Email & Phone Number

Stockholm, Stockholm County, Sweden

I lead an applied research group that uses machine learning to solve problems related to the games we make at Embark. Areas of interest include: Anti-cheat systems, physics simulation, game AI, robotics, narrative generation, speech synthesis, etc.As part of my responsibilities I:- Manage a team of smart and ambitious researchers.- Find alignment on what areas to prioritize with studio stakeholders.- Ensure that the research the team produces translates to actual improvements in the games we deliver.- Try to find the time to do the code refactors that no one else feels ownership of :)

Stockholm, Stockholm County, Sweden

In the Creative Play project at Embark we want to redefine how people create and play together. As the Craft Lead for Machine Learning within Creative Play I:- Built up a team of ML specialists, mostly in the fields of reinforcement learning and large language models.- Led the technical work for the reinforcement learning/robotics group, for both the python deep learning stack and the rust based simulation stack.- Coordinated with the leadership team and the other craft leads in matters of strategy and team organization.

Stockholm, Sweden

As a Senior Research Engineer withing machine learning I worked on applications of reinforcement learning applications to video game animation. Our specific approach is to treat animation as a robotics control problem. As part of this work I:- Implemented reinforcement learning algorithms such as soft actor critic (SAC) and proximal policy optimization (PPO) in pytorch and tensorflow.- Wrote robotics simulation systems for defining robot actuators, sensors, and rewards in Unreal Engine and Unity.- Acted as a product owner for Hive - our cloud based training cluster and the web tools we built around it.- Ran countless robotics experiments to improve gait, control, exhibit specific behaviors, etc.

Stockholm, Sweden

As a deep learning specialist at SEED I mostly worked on facial animation. We created some new methods for audio-based face animation based on Karras et al.'s work. The work resulted in two patents.I was also part of the Interim Leadership team for a short period.

Stockholm, Sweden

At Schibsted I worked mainly on their digital marketplaces such as Blocket, Finn.no and LeBonCoin. As a part of their core machine learning team I concentrated on applications of computer vision to product recommendations.

Stockholm, Sweden

For a while a friend and I tried to use deep learning and EEGs to control video games. We believed that deep learning and consumer EEGs could be the interface that would make VR take off. It didn't work out but we learned a lot.

Stockholm, Stockholm County, Sweden

As the front end lead I set the technical direction of the front end work and mentored and coached the members of the front end team.

Stockholm, Sweden

Silobreaker develops a web tool for monitoring open source intelligence sources. As a developer there I worked both on the back end in C#, developing tools such as an n-gram based language identifier and SVM based classifiers, as well as on the front end, where I created various data visualization tools and introduced new tech, such as Typescript.

Solna, Sweden

Doctor Of Philosophy (Ph.D.), Applied Mathematics

M.Sc./Civ.Ing., Engineering Physics