dr.-ing. martin franke · 2020-02-18 · deep learning recommendation systems cost-optimized...
TRANSCRIPT
Dr.-Ing. Martin Franke
Data Scientist / Solution Specialist with nearly ten years of experience in interpreting and analysing data for
driving and optimising conversions. Holding a doctoral degree in applied advanced machine learning techniques.
Proficient knowledge of statistics and analytics from computer science and psychology. Excellent understand-
ing of using state-of-the-art technology to tackle challenges and present significant solutions.
ACADEMIC QUALIFICATIONS PERSONAL DATA
Doctors Degree -
Dr.-Ing. in applied machine learning techniques
Diploma -
Computer Science with minor Psychology
Year of birth: 1988
Nationality: German
MAIN EXPERTISE LANGUAGES
Lean and appropriate data analysis for
applied automation techniques.
German (native language)
English (fluent)
Dr.-Ing. Martin Franke
TECHNOLOGIES
Programming Python, Scala/Java, SQL
Statistics / ML Python, SparkML, Pandas, scikit-learn, MXnet
Databases PostgreSQL, Google BigQuery, Amazon Redshift, MySQL, S3
Reporting Jupyter, Zeppelin, Databricks
Cloud Google Cloud, AWS, Zapier, Databricks
METHODOLOGICAL
Data analysis
Data modeling
Predictive Analytics
Real-time Analytics
Stochastic Modelling
Deep Learning
Recommendation Systems
Cost-optimized applied machine learning techniques
INDUSTRIES
Research and Development
Media and Television
Content Publisher
Ecommerce
Automotive
Dr.-Ing. Martin Franke
03/2019 - 12/2019 Automotive
ESG Mobility GmbH provides solutions for complex electronic and IT systems, especially as an automotive
supplier for the AUDI AG, Ingolstadt. The project goal is to develop a platform for automated statistical
analysis of driving behaviour regarding assistance systems and errors. To reach this, sensor data from
different BUS systems are combined with user generated data and unified in a Big Data Platform. In this
platform it is possible to detect common patterns resp. behaviour for further analysis or automated re-
actions.
Role Data Scientist
Technologies Scala, Python, Apache Spark, AWS, SparkML, MXnet
04/2018 - 02/2019 Health Care
BODYVISER GmbH provides digital, mobile solutions, to tackle chronic stress and find inner balance. For
personal assistance, it is also possible to find the right expert on the platform right away. The BODYVISER
platform serves mobile coaching, digital therapy assistance for experts and clients. Based on insights
and findings out of the therapy process, an adaptive lifestyle is established according to individualised
needs.
Role Cofounder / CTO / Data Scientist
Technologies Scala, Python, Apache Spark, AWS, SparkML, Deeplearning4j
Dr.-Ing. Martin Franke
02/2018 - 04/2018 Content Publisher / Ecommerce
Social Media Interactive GmbH with its brand BodyChange is with over 450.000 customers the top growing
online fitness brand in the DACH region. The project goal was to get a unified view of Newsletter, Shop and
so-called Coaching data. For this, a new data warehouse layer for big data sources has been accomplished.
With this data, it is now possible to answer data-centric questions such as churn rates, retention rates,
trending products, and product recommendations. As a new data-driven company, Social Media Interactive
GmbH is now capable finding best marketing strategies.
Role Data Scientist
Technologies Python, Amazon Redshift, Postgres, Zapier, Matillion ETL for Redshift
05/2017 - 12/2017 Content Publisher
Social Sweethearts Gmbh is the leading global digital publisher of individualized content. This content
consists mainly of fortune telling quizzes for Facebook users. The project goal was to optimize answers
for these quizzes by enhanced user targeting. Additionally, conversion rates, especially the share rate on
Facebook, increased by building precise user models combined with recommendation techniques. A higher
share rate, in turn, results in a higher page, and therefore, ad impressions.
Role Role: Data Scientist
Technologies Python, Google BigQuery, scikit-learn, Matillion ETL for BigQuery, Zapier
Dr.-Ing. Martin Franke
06/2016 - 04/2017 Enterprise Search
IntraFind Software AG develops products and solutions for easy searching, finding, and analyzing of
structured and unstructured information across all available data sources of a company. The project
goal was to develop methods for relevance score shaping in these search engines. In contrast to the exis-
ting approach, relevant items should be found not only by the keyword but by taking user-centric know-
ledge into account. For example, my own search history, the search history of my departement and also
company-wide announcements shape the importance of the result items.
Role Data Scientist
Technologies Python, Scala, Apache Spark, elastic Stack
07/2015 - 06/2016 Media and Television
The inovex GmbH acts as a service provider for the ProSiebenSat.1 Media SE. The goal for the longtime pro-
ject was the to migrate the data warehouse from PostgreSQL basis to the Apache Hadoop stack. To achieve
all objectives, an ETL process with more than 30 data sources, incl. Google Analytics 360, Facebook API,
Mixpanel, Webtrekk, was maintained. With this in mind, it became possible to extract and calculate busi-
ness-critical key performance indicators irrespective of data source borders.
Role Data Scientist
Technologies Python, Scala, Apache Spark, Pentaho, Apache Hadoop, PostgreSQL
Dr.-Ing. Martin Franke
09/2012 – 04/2016 Research / Health
Regular daily routines not only increase the overall well-being but also affect peoples’ health and the
perceived stress level. Large discrepancies in these rhythms may result in sleep disorder up to chronic
depressions. Nowadays, daily routines can only be manually supervised by experts in order to avoid these
negative consequences. The intention of the proposed solution is to offer automated and preventive assis-
tance for this complex process. This is done, by combining several sources of information, e.g. domestic
sensor values, smartphone data, social networks, fitness trackers, to determine daily activities with
machine learning techniques. In the end, calculated routines from daily activities result in patterns, out-
liers, and corresponding causes.
RolE Research Associate
Technologies Java, Python, WEKA, nodejs, OWL