Stefan is available for hire

Stefan Decker

Verified Expert in Engineering

Data Scientist Developer

Location

Berlin, Germany

Toptal Member Since

February 22, 2021

Stefan is a data scientist and entrepreneur who has worked on projects ranging from document processing, sentiment analysis to time-series classification. He co-founded multiple companies and has a passion for defining data science projects with non-technical colleagues, working on the entire project pipeline, goal definition, data retrieval & cleaning, modeling, and deployment. He raised funds, pivoted, and guided negotiations to the successful acquisition of his companies.

Machine Learning Artificial Intelligence (AI)Pandas NumPy Scikit-learn Natural Language Processing (NLP)Deep Learning Google Sheets Python PostgreSQL TensorFlow Web Scraping SQL Git Amazon Web Services (AWS)

Portfolio

Mathemathicai

Python, Pandas, NumPy, Scikit-learn, TensorFlow, Amazon Web Services (AWS)...

Zeitgold

Python, PostgreSQL, NumPy, Scikit-learn, TensorFlow, Django, Pandas...

I2x

Git, NumPy, Pandas, TensorFlow, Python, PostgreSQL, Scikit-learn...

Experience

Machine Learning - 6 years NumPy - 4 years Generative Pre-trained Transformers (GPT) - 4 years Python - 4 years Pandas - 4 years Natural Language Processing (NLP) - 4 years Scikit-learn - 4 years PostgreSQL - 3 years

Availability

Part-time

Preferred Environment

PyCharm, MacOS, Slack

The most amazing...

...product I've developed was a booking automation model. I drove conceptualization, data retrieval, modeling, and deployment.

Work Experience

Freelance Data Scientist

2020 - PRESENT

Mathemathicai

Created a BERT model for the sentiment classification of English text.
Conceptualized and trained a model for the IoT classification task based on power consumption.
Supported with iterative model improvements by creating custom evaluation pipelines.

Technologies: Python, Pandas, NumPy, Scikit-learn, TensorFlow, Amazon Web Services (AWS), Google Cloud, Deep Learning, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Machine Learning, Artificial Intelligence (AI)

Data Scientist

2019 - 2020

Zeitgold

Conceptualized, trained, and implemented deep learning and other machine learning models for document classification to significantly reduce the need for manual labor.
Provided analysis and took part in discussions with top-level executives about the machine learning strategy of the company.
Improved and maintained existing models and integrated retraining functionality in the core back end.

Technologies: Python, PostgreSQL, NumPy, Scikit-learn, TensorFlow, Django, Pandas, Amazon Web Services (AWS), Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Deep Learning, Machine Learning, Artificial Intelligence (AI)

Data Scientist

2017 - 2019

I2x

Built production-ready NLP features (Python, C++) that are now an integral part of the product.
Trained custom word vector models on movie subtitles to improve the handling of swear words.
Led a team of two machine learning engineers through the development of critical NLP components.
Acted as technical project lead for a machine learning project with the largest customer.
Conceptualized the data labeling process and built classifiers to improve labeling efficiency by using active learning.

Technologies: Git, NumPy, Pandas, TensorFlow, Python, PostgreSQL, Scikit-learn, Generative Pre-trained Transformers (GPT), Natural Language Processing (NLP), Machine Learning, Artificial Intelligence (AI), Leadership

Co-founder and Managing Director

2014 - 2017

Invincible Brands

Bootstrapped brand with no external funding to €20,000 monthly revenue in less than 12 months.
Built an Instagram scraping and analytics tool using Excel (XML Import) that enabled rapid scaling of marketing activities.
Initiated and facilitated the acquisition by Invincible Brands.
Set up a process to outsource almost all operational work.
Hired and managed a team of four fulfillment and customer support employees.

Technologies: Web Scraping, Leadership

Growth & Analytics Specialist

2015 - 2016

SwitchUp

Set up a complex rule-based Excel tool to manage the send out of reminder emails for customers. This made the development of a Ruby application temporarily unnecessary.
Developed an Excel BI tool using system and Mixpanel data to monitor company-wide KPIs and support strategic decisions.
Set up and managed A/B testing campaigns to help the company find product-market fit.

Technologies: Mixpanel, Facebook Ads, Google Sheets

Co-founder and CMO

2012 - 2014

JUNIQE.com

Raised €450,000 from institutional investors. Co-created the pitch deck and financial model and took part in the negotiations.
Led a cross-functional team of seven, including designers, backend- and frontend developers, to accomplish a product pivot.
Set up the web and mobile BI system using Google Analytics, Mixpanel, and adjust.io.

Technologies: Google Ads, Facebook Ads, Google Analytics, Mixpanel, Leadership

Consultant

2011 - 2012

EY

Supported several M&A deals including the biggest real estate deal in Germany since 2008.
Calculated ß factors with regression by using historical stock prices for all internal customers of Ernst & Young Germany.
Managed the outsourcing of data retrieval tasks in an M&A project to meet critical deadlines.

Technologies: Accounting, Financial Modeling, Capital Markets, Consulting

Experience

Deep House Spy

https://github.com/sbadecker/deep_house_spy

The Deep House Spy is an artist classification project based on a convolutional neural network (CNN). CNNs have proven to be really effective for image classification and can also work with audio when put in the right form.

The idea project came when I listened to a deep house DJ set on Soundcloud. I really liked the song that I was listening to, but Shazam just wouldn't recognize it. This happens very often and the reason for this is (most of the time) that Shazam doesn't have the song in its database because it has not yet been released.

My approach to solving this problem was to learn the style of artists by using songs that have been released already and then identifying the respective artists of the unreleased songs.

Education

2004 - 2010

Master's Degree in Business Administration

Georg-August-Universität - Göttingen, Germany

Certifications

MARCH 2018 - PRESENT

Deep Learning Specialization

Coursera

APRIL 2017 - PRESENT

Data Science

Galvanize

Skills

Libraries/APIs

Pandas, NumPy, Scikit-learn, TensorFlow

Tools

Google Sheets, PyCharm, Google Analytics, Git

Languages

Python, SQL

Paradigms

Data Science

Platforms

Mixpanel, Amazon Web Services (AWS)

Storage

PostgreSQL, Google Cloud

Frameworks

Django

Other

Machine Learning, Artificial Intelligence (AI), Deep Learning, Natural Language Processing (NLP), Generative Pre-trained Transformers (GPT), Facebook Ads, Google Ads, Web Scraping, Capital Markets, Leadership, Consulting

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring