freiberufler PhD CS, Data Science, ML Engineer, Deep Learning auf freelance.de

PhD CS, Data Science, ML Engineer, Deep Learning

zuletzt online vor wenigen Stunden
  • auf Anfrage
  • 64390 Erzhausen, Hessen
  • auf Anfrage
  • de  |  en
  • 08.04.2024

Kurzvorstellung

Computer Scientist, Data Science, Machine Learning, Deep Learning, Computer Vision, NLP, Scientific Visualization, Information Visualization, Project Management, Team Lead

Auszug Referenzen (2)

"Björn hat uns von Anfang an und bis Projektende durch umfassendes Wissen, Organisationstalent, Kundenfokus und Menschenkenntnis begeistert."
Product Owner / Tech Lead Data Science Platform
Michael Cornel
Tätigkeitszeitraum

2/2021 – 12/2023

Tätigkeitsbeschreibung

o Build the Team, Define Vision and roadmap around the main persona Data Scientist
o Iterate on minimum-viable product (my chosen approach)
o Technical and MLOps aspects of operating hardware/infrastructure, edge-cloud, cloud, and also in development context with CI/CD, considering latency and throughput SLAs required in industrial context
o Participate in concrete Machine Learning projects to implement solution based on the DSP. Mostly in industrial context and research, including monitoring for data drift and general reliability of deployed models
o Most notable a project around our Federated Learning Feature to classify 3D scan for Alzheimer detection in cooperation with Fraunhofer, presented at HannoverMesse, , AI-NET, and eucnc.eu

As edge-first product resource consumption and sharing of GPU resources was a priority. The DSP features we implemented address all aspects of data processing from ingestion in heterogeneous IT landscapes, clean-up and transformation, all the way to hosting and deploying ML models in a privacy preserving fashion. Our unique approach around copy-on-write-snaphots and processing pipelines allowed for Reproducible ML in all workflows, without any lock-in.

Eingesetzte Qualifikationen

Projektleitung / Teamleitung (IT), Continuous Delivery (CDE), DevOps (allg.), ETL, Git, Jira, Kotlin, Maschinelles Lernen, Neuronale Netze, Python, PyTorch, TensorFlow

"B. hat immer eine Lösung gefunden, die uns als Team und unser Produkt besser gemacht hat."
Data Science Coach, Machine Learning Engineer
Michael Rohrmann
Tätigkeitszeitraum

6/2016 – 6/2019

Tätigkeitsbeschreibung

Building Data Science Team for in-house DS department
• Inception and implementation of strategy to begin and iterate DS
projects to maturity in an agile fashion, including getting buy-in
from leadership...
• Coaching new Data Scientist in finding and implementing
appropriate project solutions
Later Product Owner for several projects driven by that strategy and selection from demand funnel with key account manager.

Hands-on implementations
• Implementations for several DS projects in that setting
with python Implementations for computer vision, natural language processing (PDF docs), text entity recognition, table extraction (GPU accelerated Yolo and Faster RCNN), text summarization, sentiment analysis
• Innovative oneshot Siamese CNN solution using keras (tensorflow) for customizable glyph extraction (think checkboxes in scanned docs). Also using OpenCV, pandas, numpy
• Innovative Image Keypoint Detection (like SIFT but for text images) for dewarping scanned Text documents. Using OpenCV, Stable for translation, scaling and rotation
• ID Sequence alignment using Needleman–Wunsch algo
• Sequence Pattern recognition in 2D floating point time series
using sklearn support vector machine (SVM) with sliding
window
• One-off efficient Labeling web-application that saved project only
feasible with supervised learning approach, using flask, nginx,
dash
• Collaboration with DevOps team to create SoA, to provide continuous integration (CI), UI, hosting and abstraction framework for all PwC employees to allow implementation of their own data driven solutions, writing only python code. Kafka for high throughput data delivery, kubernetes and docker for modules

Eingesetzte Qualifikationen

Lean Startup, Maschinelles Lernen, Projektleitung / Teamleitung (IT), Python, SOA (Serviceorientierte Architektur)

Qualifikationen

  • Big Data
  • Computer Vision
  • Data Science
  • Deep Learning
  • Maschinelles Lernen
  • Natural Language Processing (NLP)
  • Projektleitung / Teamleitung (IT)
  • Python
  • Software Architektur / Modellierung
  • Solution Architektur

Projekt‐ & Berufserfahrung

Product Owner / Tech Lead Data Science Platform
German Edge Cloud, Eschborn
2/2021 – 12/2023 (2 Jahre, 11 Monate)
IT & Entwicklung
Tätigkeitszeitraum

2/2021 – 12/2023

Tätigkeitsbeschreibung

o Build the Team, Define Vision and roadmap around the main persona Data Scientist
o Iterate on minimum-viable product (my chosen approach)
o Technical and MLOps aspects of operating hardware/infrastructure, edge-cloud, cloud, and also in development context with CI/CD, considering latency and throughput SLAs required in industrial context
o Participate in concrete Machine Learning projects to implement solution based on the DSP. Mostly in industrial context and research, including monitoring for data drift and general reliability of deployed models
o Most notable a project around our Federated Learning Feature to classify 3D scan for Alzheimer detection in cooperation with Fraunhofer, presented at HannoverMesse, , AI-NET, and eucnc.eu

As edge-first product resource consumption and sharing of GPU resources was a priority. The DSP features we implemented address all aspects of data processing from ingestion in heterogeneous IT landscapes, clean-up and transformation, all the way to hosting and deploying ML models in a privacy preserving fashion. Our unique approach around copy-on-write-snaphots and processing pipelines allowed for Reproducible ML in all workflows, without any lock-in.

Eingesetzte Qualifikationen

Projektleitung / Teamleitung (IT), Continuous Delivery (CDE), DevOps (allg.), ETL, Git, Jira, Kotlin, Maschinelles Lernen, Neuronale Netze, Python, PyTorch, TensorFlow

Founding CEO (Festanstellung)
efabu GmbH, Darmstadt
7/2017 – 7/2020 (3 Jahre, 1 Monat)
IT & Entwicklung
Tätigkeitszeitraum

7/2017 – 7/2020

Tätigkeitsbeschreibung

Building a Speech Recognition based solution for tax ready drivers log

• Agile, Lean startup based approach around actionable metrics
• Speech Recognition based solution for tax ready drivers log (German tax "Fahrtenbuch") initially with modified one-function-android prototype with GPS tracking and voice recording, later OBD Hardware for car GPS tracker and iPhone app
• Cloud Service for streams of data frames from ioT device (in-car tracker with GPS and milage via OBD2), and voice recordings from mobile app
• Custom Speech Recognition (personal names, in-car noise,...) and dashboard for tax related QA

Eingesetzte Qualifikationen

Big Data, Spracherkennung, Agile Entwicklung, Software Architektur / Modellierung, Amazon Web Services (AWS), Internet of Things (IoT), Projektmanagement

Data Science Coach, Machine Learning Engineer
PriceWaterhouseCoopers, Frankfurt
6/2016 – 6/2019 (3 Jahre, 1 Monat)
Unternehmensberatung
Tätigkeitszeitraum

6/2016 – 6/2019

Tätigkeitsbeschreibung

Building Data Science Team for in-house DS department
• Inception and implementation of strategy to begin and iterate DS
projects to maturity in an agile fashion, including getting buy-in
from leadership...
• Coaching new Data Scientist in finding and implementing
appropriate project solutions
Later Product Owner for several projects driven by that strategy and selection from demand funnel with key account manager.

Hands-on implementations
• Implementations for several DS projects in that setting
with python Implementations for computer vision, natural language processing (PDF docs), text entity recognition, table extraction (GPU accelerated Yolo and Faster RCNN), text summarization, sentiment analysis
• Innovative oneshot Siamese CNN solution using keras (tensorflow) for customizable glyph extraction (think checkboxes in scanned docs). Also using OpenCV, pandas, numpy
• Innovative Image Keypoint Detection (like SIFT but for text images) for dewarping scanned Text documents. Using OpenCV, Stable for translation, scaling and rotation
• ID Sequence alignment using Needleman–Wunsch algo
• Sequence Pattern recognition in 2D floating point time series
using sklearn support vector machine (SVM) with sliding
window
• One-off efficient Labeling web-application that saved project only
feasible with supervised learning approach, using flask, nginx,
dash
• Collaboration with DevOps team to create SoA, to provide continuous integration (CI), UI, hosting and abstraction framework for all PwC employees to allow implementation of their own data driven solutions, writing only python code. Kafka for high throughput data delivery, kubernetes and docker for modules

Eingesetzte Qualifikationen

Lean Startup, Maschinelles Lernen, Projektleitung / Teamleitung (IT), Python, SOA (Serviceorientierte Architektur)

Lead Engineer (Festanstellung)
Nokia Here GmbH, Schwalbach bei Frankfurt
10/2010 – 5/2015 (4 Jahre, 8 Monate)
IT & Entwicklung
Tätigkeitszeitraum

10/2010 – 5/2015

Tätigkeitsbeschreibung

(Nokia Location & Commerce), Geographic Information Systems
Lead Engineer and Senior Engineer, Project Management (fulltime employee)
• 3 years team lead for 5 developers, Nokia HERE Geocoding Service (NGC) improvements
Coordination between devs, management, data analysts, testers
Work package definition and prioritization, QA, Code reviews
• Software Development for NGC, 100M requests/day (lucene and proprietary index technology, Java)
• NGC Machine Learning components for query classification (Hidden Markov Model, decision trees)
• Massive parallel data analysis (Hadoop Cloud) applying Machine Learning and statistical approaches to user request analysis for NGC quality improvement (Random Forest, Naive Bayes classifier)
• Project Lead for NGC quality improvements (country Focus)
• Training class instructor for Geodata processing
• Job postings and interviews, onboarding, University liaison (MA thesis,...)

Eingesetzte Qualifikationen

Big Data, Maschinelles Lernen, Projektleitung / Teamleitung (IT), J2EE (Java EE), Software engineering / -technik, Amazon Web Services (AWS)

Senior Software Engineer, Machine Learning Engineer, Project Management
United European Gastroenterology Foundation, Remote
5/2010 – 10/2010 (6 Monate)
Gesundheitswesen
Tätigkeitszeitraum

5/2010 – 10/2010

Tätigkeitsbeschreibung

• Lucene/Solr based search engine for multimedia conference documents (videos, presentation slides, PDFs), portal integration (E-Learning)
• Taxonomy Definition and population using K-Means clustering (Apache Mahout) to categorize conference documents
• Taxonomy for faceted browsing for interactive exploration and refinement

Eingesetzte Qualifikationen

Maschinelles Lernen, Java (allg.)

Researcher Associate, Software-Architect, Project Management (Festanstellung)
FernUniversität in Hagen, Remote, Frankfurt
5/2008 – 4/2010 (2 Jahre)
Hochschulen und Forschungseinrichtungen
Tätigkeitszeitraum

5/2008 – 4/2010

Tätigkeitsbeschreibung

• Lecturer Information Visualization class, supervisor for master theses, lab assignments and research assignments
• EU-Project ParseInsight project management
• ParseInsight GapAnalysisTool: Dynamic visualization of tree structures in support of methods for analysis and filtering of survey results (Java, Groovy)
• Real-time ridesharing mockup using PTV MobilityPlatform POIAdministration and Routing Service for ADAC. Implementation of Matching (rider to hitcher) component (Java)
• […]: GIS-Module for geographic meta-search engine for carsharing portals: -Hyperlink entfernt- (PHP, PostGis)
• JustGetThere: Mobile Android client for Hopping.de web service with route visualization
• RouteVis3D: Geographic Information System for querying, visualization and manipulation of traffic and route data in systems like Hopping.de. (Java, Java3D, Swing, PostGis)
• OSMRouting: OpenStreetMap based routing module (Java, PostGis)
• OSMConnect: Quality analysis of OpenStreetMap data regarding routing
• IcoPort: Matching and visualization of possibilities for cooperation of delivery services, based on delivery routes and districts for logistics fleet management (PHP, PostGis)
• RideSim: Simulation analysis and visualization of real-time ridesharing systems. Cooperation with TU-Darmstadt and department of transportation/ Hessischem Landesamt für Straßen- und Verkehrswesen (Java, PostGis, Java Topology Suite)
• MapCube: 3D Focus+Context visualization technique for geographic information spaces (Java, JOGL, OpenGL ES) as well as proof of concept for porting to iPhone (C++, OpenGL ES).
• ParseInsight Interactive Map: Google maps based visualization and search of location based project data (PHP, Javascript, Google Maps)
• Consulting myTellus.com (Travel portal; Swedish Centre for International Youth Exchange) regarding integration of GIS visualization- and server components

Eingesetzte Qualifikationen

Grundlagenforschung, Projektleitung / Teamleitung (IT), Software engineering / -technik

Scientific Visualization, Software Engineer
Carl Zeiss Vision, Aalen
5/2006 – 4/2008 (2 Jahre)
Fertigungsindustrie
Tätigkeitszeitraum

5/2006 – 4/2008

Tätigkeitsbeschreibung

Scientific 3D visualization for java applet B2B shop system (LogOn): Mapping of Zeiss lens data to parametric 3D model and tesselation for visualization of glass lenses (Java3D). OpenGL diagnostic tool (C++). LogOn has 15000 users.
• Web services to make prescription calculation logic available to server side business logic. Access .Net and Java web services to integrate wavefront measuring device into LogOn client (Axis, JAX-WS, SOAP)
• Performed Code Review and provide marked up copies to team (8 developers) to improve quality, in particular maintainability of the LogOn system.
• Introduced structured test procedure and test automation for front and back-end components
• Refactoring and redesign of 3-tier architecture of LogOn to improve performance (J2SE, J2EE, JBoss, XML/XSL)

Eingesetzte Qualifikationen

Software engineering / -technik

Researcher, Diploma Thesis
Tokyo University, (Robotics) Department of Informa, Tokyo
10/2004 – 10/2005 (1 Jahr, 1 Monat)
Hochschulen und Forschungseinrichtungen
Tätigkeitszeitraum

10/2004 – 10/2005

Tätigkeitsbeschreibung

Embedded Development: Implementation of high-speed eye-tracking to capture saccadic eye movements
Implementation of image processing algorithms for Windows and special purpose hardware (Vision Chip)
Keil 8051 C51, C++

Eingesetzte Qualifikationen

Embedded Entwicklung / hardwarenahe Entwicklung

Senior Software Engineer
International SoftDevices Corp., Remote
11/2001 – 4/2006 (4 Jahre, 6 Monate)
IT & Entwicklung
Tätigkeitszeitraum

11/2001 – 4/2006

Tätigkeitsbeschreibung

Continued development and extension of my previous projects and integration with new projects, as well as implementation of new SDServices, troubleshooting and bug fixes (Java, C++)
• Introduced transaction based multithreading model for SDServices (C++)
• GUI extensions for myMissions client, e.g., chat and whiteboard (Java,
Swing)
• Embedded development: Integration of Arula-Box as SDService (8051 C/SDCC small device c compiler)

Eingesetzte Qualifikationen

Software engineering / -technik

Senior Software Engineer, -Architect, Team Lead (Festanstellung)
International SoftDevices Corp., Santa Clara, CA, USA
11/1999 – 10/2001 (2 Jahre)
IT & Entwicklung
Tätigkeitszeitraum

11/1999 – 10/2001

Tätigkeitsbeschreibung

Participated in developing distributed system for control and execution of user defined business processes (SDPlatform) based on default and user defined SDProcesses (C++)
• As Lead Senior Software Engineer responsible for requirements analysis and specification, design specification, cost and time estimates, as well as implementation of RecoverSystem, subsystem guaranteeing high availability of SDProcess states (C++)
• Embedded Development: PalmOS Clients for SDPlatform (C++ PalmOS 3.5)

Eingesetzte Qualifikationen

Software engineering / -technik

Software Engineer
Fraunhofer Institute for Computer Graphics Researc, Darmstadt
6/1998 – 9/1999 (1 Jahr, 4 Monate)
IT & Entwicklung
Tätigkeitszeitraum

6/1998 – 9/1999

Tätigkeitsbeschreibung

Contributed to development of „genesis“, a radiosity based 3D rendering system for 5-sided CAVE (immersive virtual reality environment) (C/C++, IrixGL)
• Usability and 3D rendering improvements (C, motif)
• Pre-processing of very large3D models (e.g. Mercedes Benz car models, Frankfurt airport model for Fraport AG) to enable radiosity illumination in genesis

Eingesetzte Qualifikationen

Grafiksoftware (allg.), Objektorientierte Programmierung (OOP), CAE (rechnergestützte Entwicklung / computer-aided engineering)

Software Engineer
Hoffmann Versicherungen, Langen, Hessen
10/1996 – 9/1998 (2 Jahre)
Versicherungen
Tätigkeitszeitraum

10/1996 – 9/1998

Tätigkeitsbeschreibung

Requirement analysis and implementation of InsuranceCalculator for calculation and comparison of insurance rates (Visual Basic)
• Design of questionnaire to gather requirements from insurance agents

Eingesetzte Qualifikationen

Objektorientierte Programmierung (OOP)

Software Engineer
Mannesmann Dematic, Offenbach
10/1996 – 9/1998 (2 Jahre)
High-Tech- und Elektroindustrie
Tätigkeitszeitraum

10/1996 – 9/1998

Tätigkeitsbeschreibung

Implementation of proprietary printer driver for data warehouse system (C/C++, driver programming), as well as data conversion, import/export routines, etc.
• Improving the warehouse inventory software, as well as the end-user software. This involved customization of SAP solutions using ABAP, in particular, mapping the kanban approach of the proprietary solution to SAP functionality.

Eingesetzte Qualifikationen

Software engineering / -technik

Zertifikate

Promotionsurkunde
2015

Ausbildung

Informatik
Dipl. Informatiker
2005
Darmstadt

Über mich

I still love writing code. Climbing around the organizational hierarchy with my roles has always been inspired by delivering business value, as well as striving for elegance in technical things. Seeing the world from these perspective has always given me an edge and shown me that they can be in alignment, and if they do, truly great solutions for people can be built in software. Architecting and coding understandable, scalable and maintainable systems follows naturally.
Building solutions that way has changed a lot over the years, regarding technologies. Not always easier, but definitely more awesome. Massive streams of data from IoT devices along with mobile apps being listened to by ML algorithms to deliver those solutions. Embedded into the user's reality, and using the power of the cloud. That is where I want to be.

It would be my pleasure to help you in building those solutions. I am looking forward to it. Preferably from my home office in the near future :-)

Weitere Kenntnisse

• Seasoned team lead, software architect and -engineer in international (English) projects, 1999 two years fulltime in Silicon Valley startup.
• PhD in Computer Science with emphasis on GPU accelerated 3D graphics and Information Visualization
• Skills and Certifications in Data Science & Artificial Intelligence

• Recently: 3 years Data Science (DS) Team Coach at PriceWaterhouseCoopers
- Building Data Science Team for inhouse DS department
- Machine Learning and AI consulting and implementations
- Computer Vision, Natural Language Processing
- Microservice architecture for Data Insight Platform

• Until 2015: Lead Engineer at Nokia Here with emphasis on
- Data Analytics in Big Data
- Machine Learning
- Architecting improvements for Nokia Here Geocoder (NGC)

• Embedded/IoT: High Speed Eye Tracker, RePhone, 8051, Arduino, Raspberry Pi

=== IN DETAIL ===
MANAGEMENT:
Project and Team lead, CEO role
Always agile since 1999, even without manifesto
Lean startup at heart

TECH:
Data Science & Artificial Intelligence:
• Most work in python, a little R
• Coursera Johns Hopkins University, Data Science Specialization (R)
• Coursera deeplearning.ai, Deep Learning Specialization (python)
• Python, keras, tensorflow, NLTK, spacy, gensim, sklearn, OpenCV, pandas, numpy
• Sequence models, Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU)
• Deep learning for computer vision with Convolutional neural networks (CNN), Yolo, Faster R-CNN
• Semantic search with word2Vec
• Support Vector Machines, k-means, Decision Trees, Classifiers
• Hidden Markov Models in HERE Geocoder (100M requests/day)
• Big Data Random Forest and Naive Bayes for Nokia Geocoder query-log based quality improvement
• Neural networks and genetic algorithms
• Information retrieval, TFID

Agile Software Development Methodologies:
• 5 years Scrum at Nokia,
Daily standups as participant and Scrum master, Scrum-of-Scrums, kanban approaches
Responsible for backlog grooming in lieu with PO
Product Owner Training and PO for 2 projects over 3 years
• 3 years Scrum at PricewaterhouseCoopers, including as PO

More recent additions
• Kubernetes, docker
• Kafka, RabbitMQ, redis
• python flask, dash
• Azure CosmosDB
• Neo4j
• Mobile apps for Android, iOS: Flutter with dart, swift


Ancient
• UML, Enterprise Architect
• Java (J2EE, Swing, JavaFX, Java3D, Netzwerk, JNI, RMI)
• Python
• Groovy
• Pig / hadoop
• R
• Eclipse IDE, IntelliJ IDE, MS Visual Studio, VS Code
• C/C++
• C#
• Embedded-Development (MS embedded VC++, Keil C51, sdcc)
• Metrowerks Codewarrior for PalmOS (C++)
• 3D Visualization with OpenGL, OpenGL ES, Java3D

• Mobile: Flutter with dart
• XML/XSL, Axis, JAX-WS
• Turbo Pascal, Delphi
• Perl
• Tcl/Tk
• Lisp
• Visual Basic
• ABAP (SAP)
• Shell scripts
• SQL
• Web: HTML, Javascript (a little CSS)
• TestComplete, SoapUI
• SOAP, JAXB
• JBoss, tomcat, spring
• CVS, Subversion, Git, Maven, Gradle, Bamboo,
• Crucible code reviews




Operating Systems:
• Windows
• Unix (Suse, RedHat, Ubuntu, HP-UX)
• OSX
• Android, iOS

Databases and Infrastructure:
• Postgres (PostGis)
• MongoDB, Azure CosmosDB
• Neo4j
• MySQL
• Oracle
• Access

Miscellaneaous:
• GUI design, web-development, multimedia data formats, data retrieval, Bluetooth
• Administration of Windows and Unix, Firewall, Router, Http-Proxy, networks

Persönliche Daten

Sprache
  • Deutsch (Muttersprache)
  • Englisch (Fließend)
Reisebereitschaft
auf Anfrage
Arbeitserlaubnis
  • Europäische Union
Home-Office
bevorzugt
Profilaufrufe
173
Alter
48
Berufserfahrung
27 Jahre und 6 Monate (seit 10/1996)
Projektleitung
13 Jahre

Kontaktdaten

Nur registrierte PREMIUM-Mitglieder von freelance.de können Kontaktdaten einsehen.

Jetzt Mitglied werden