Machine Learning Working Student - Berlin, Deutschland - dida Datenschmiede GmbH

dida Datenschmiede GmbH

Geprüftes Unternehmen

Berlin, Deutschland

vor 3 Wochen

Geschrieben von:

Lena Wagner

beBee Recruiter

Beschreibung

dida is a machine learning software company with exciting problems for instance in computer vision and natural language processing.

Our team tackles applied problems for different customers by using latest scientific advancements (especially in deep learning) and therefore believes that research oriented thinking can help solving real-world problems more efficiently.

You are
- studying mathematics or physics
- interested in applied statistics and in programming with Python, Julia or something alike
- eager to learn more about modern machine learning approaches such as deep learning
- planning to stay in Berlin at least 1.5 more years and to work 20h/week
- interested in solving real-world problems with creative ideas

You will
- meet an interdisciplinary team of people with a solid background in mathematics and statistics
- have flexible working hours and a nice office with good coffee in Berlin Schöneberg (of course, you can work mostly remote during Corona)
- work for 1.5-2 years on an interesting image segmentation task; it's about detecting certain landscape patterns in satellite imagery
- be part of a project team of 3 experienced machine learning scientists, who will ensure that you learn a lot (please note: about 25% of your time you will need to do data annotation, but the other 75% are pure programming/machine learning in Python)

Find below a short description of two of our current projects.

Estimate the amount of solar panels that fit on a roof (computer vision):

Given a satellite picture and a ground image of a house, automatically detect certain elements of a roof (including obstacles, dormers etc.) in order to find out how many solar panels fit on it.

This involves inferring 3d information from 2d pictures in order to infer the roof pitch.

Detect, classify and suggest legal effectiveness of text paragraphs (NLP):
Automatically go through thousands of legal documents with the goal to classify dedicated paragraphs and check their legal effectiveness.

This involves converting scans to text, coming up with a labelling scheme (problem modelling), and detecting different paragraphs automatically, before tackling the inference task.

Art der Stelle:
Teilzeit, Werkvertrag

Arbeitsstunden: 20 pro Woche

Gehalt: 13,00€ pro Stunde

Arbeitsort:
Zum Teil im Homeoffice in 10827 Berlin