ABOUT

picture of myself

Dr. Katharina Rasch

Data scientist | computer vision engineer | teacher
PhD Computer Science (KTH Stockholm)
Freelancer in Berlin → Work with me

hello@krasch.io | github | twitter

PROJECTS

TALKS

WORK WITH ME

Available for freelance projects in Berlin and remote

What I can help you with:

Building machine learning products
In my 15 years as a software engineer, I have had the chance to work with a lot of different technologies. As such, I can help you build products across the full stack: machine learning / computer vision (scikit-learn, pytorch, tensorflow, keras), backend (Python), frontend (HTML, CSS, Javascript), cloud deployment (AWS, azure). Can quickly get up-to-speed with technologies not listed here.
Software engineering best practices in data science
I strongly believe that data science and machine learning benefit from practices already established in software engineering. Unit/integration tests help us build more robust products, continuous integration and deployment help avoid the "the notebook only runs on my machine" syndrome, code review helps teams build common knowledge about the data and its intricacies.
Training data acquisition
In most machine learning projects, getting more/better training data beats spending time on improving model architectures and hyperparameters. I have worked with in-house data collection/annotation (using standard + custom tooling), with crowd-sourcing and with synthetic image generation, and can help you design and execute your data acquisition strategy.
Mentoring
I have several years of teaching / mentoring experience, including a hands-on data science lab, that so far over 100 university+private students have taken. I would be happy to mentor any junior data scientists on your team in their day-to-day work. Also open to individual mentoring requests.

Contact

If you are interested in working with me, just send a quick email to hello@krasch.io.
I'll be in touch to schedule an intro meeting!


Data privacy: anonymous website usage statistics are collected using https://plausible.io/