Georg Walther

http://georg.io http://uk.linkedin.com/in/georgwalther
contact@georg.io http://careers.stackoverflow.com/georgwalther

Summary

I am a biochemist & computational biologist by training, have worked as a software engineer for a startup, and am now in training to become a data scientist.

On the technical side I am enthused by data analysis, predictive analytics, and engineering clean code. I am also enthusiastic about understanding the business opportunities that data-driven decision making can produce.

In my work approach I am proactive and enjoy both autonomous and collaborative work.


Work Experience

Data Scientist at Singular Intelligence (2014), London, United Kingdom

As part of the Science 2 Data Science summer school I did work for Singular Intelligence, a startup that helps businesses develop data- and analytics-driven corporate strategies.

As part of my work I did data exploration and visualization in Python with pandas and matplotlib. I further employed scikit-learn (linear models and random forests) to develop a predictive model of sales performance as a function of business-relevant features.

Software Engineer at WalletSaver (2014), Rome, Italy

WalletSaver is a startup in the space of personalized and data-driven recommendation systems. WalletSaver develop an app that measures securely parameters of mobile phone usage and recommends the best mobile phone plan based on cost, signal coverage, and phone carriers called.

I engineered the recommendation engine behind WalletSaver in Python with SQLAlchemy (a SQL object-relation mapper library) and Memcached to speed up data queries. I followed a test-driven development methodology for this project and performed timing-based optimization such as precomputing certain data. I sought actively customer feedback and amended the recommendation engine in response to feedback.

I also engineered and maintained geolocation code, critical for signal coverage analysis, in Python with the requests library.

Teaching Assistant at ETH Zurich (2008 - 2009), Zurich, Switzerland

Research and preparation of lecture notes, preparation of exercise questions, preparation of exam questions, grading exams.


Education and Certificates

Science 2 Data Science (August 2014 - September 2014), London, United Kingdom

Science 2 Data Science (S2DS) is an industry-sponsored summer school that leads graduates with numerical backgrounds into the field of data science.

At this course I learned about Hadoop, SQL databases, NoSQL databases, natural language processing (NLP), statistics with R, machine learning with scikit-learn, and data visualization.

I further learned about business opportunities that arise from data-enhanced and data-driven business models, economics, marketing, finance, project management, and corporate strategy.

I summarize my experience at this school at http://georg.io/s2ds.

PhD Researcher (September 2010 - September 2014), John Innes Centre, Norwich, United Kingdom

My research evolved around modeling cell polarity under a number of complex conditions. The formation of poles (e.g. front v. back) is important for cells to know which direction to grow or migrate. Specifically, I studied these equations in extreme environments such as exceptionally large and small cells.

I wrote unit tested code in Python with NetworkX that implements a mathematical framework for bifurcation analysis of large reaction networks. Here I had to rephrase a number of combinatorial problems as classical graph-theoretic problems to reduce computational cost. One particular graph-theoretic algorithm, the enumeration of all cliques, was not available in NetworkX at the time. I implemented a published algorithm for this problem and my implementation has since been added to NetworkX.

I further implemented variants of Crank-Nicolson in Python with NumPy to integrate numerically my specific convection-diffusion-reaction systems. I performed timing-supported optimization of my code and added inline C code where necessary for further speed improvement.

During my PhD I also wrote numerical simulations of a high-dimensional dynamical system in pure C with support of the GNU Scientific Library.

Visiting Scholar (2009 - 2010), The University of British Columbia, Vancouver, Canada

Research and write up of Master thesis: Stochastic Simulations of Cell Polarization and Wave Pinning.

For this project I implemented Gillespie’s stochastic simulation algorithm where I wrote the simulation setup in MATLAB and time-critical code in C. For the same project I made extensive use of numerical bifurcation packages such as XPP/auto and MatCont.

Master of Science (2010), ETH Zurich, Zurich, Switzerland

Self-designed major in computational biology and biological chemistry with strong focus on numerical analysis and data analytics.

Specific mathematical and computational courses that were part of my curriculum:

Bachelor of Science (2008), ETH Zurich, Switzerland

Major in biochemistry.


Skills

Type Name Skills
Programming Python Advanced, numerical simulations, database-driven backend development, data analysis, predictive analytics (research projects and work for startups)
C Advanced, numerical simulations, stochastic simulations (research projects)
C++ Familiar and learning (education and self-taught)
Operating Systems Linux, Windows Advanced user, Python package development and distribution, scripting
Databases PostgreSQL, MySQL Working knowledge, data modeling, querying, and data integration into time-critical code (work for a startup)
Neo4j, Rexster Working knowledge (personal side projects)
Project Management and Collaboration git Advanced, have worked on shared codebase in a startup
Continuous Integration Working knowledge, have used Jekyll CI to test academic code on Linux and OSX
Test-Driven Development Working knowledge, I have used TDD in a startup setting
Communication Work communication I am comfortable presenting my work in informal and formal settings
Scientific writing I have authored two scientific articles as lead author
Technical writing I write technical articles about my research and side projects at http://georg.io

Publications and Software

GraTeLPy: Graph-Theoretic Linear Stability Analysis

Walther GR, Hartley M, Mincheva M (2014): GraTeLPy: Graph-Theoretic Linear Stability Analysis, BMC Systems Biology (software available at http://github.com/gratelpy)

I implemented a graph-theoretic framework for the analysis of dynamical systems in Python with NetworkX. I designed the algorithm, implemented it along with auxiliary methods that were necessary, and combined graph-theoretic approaches in novel ways to reduce combinatorial blowup.

I further responded to user feedback with extensive testing on Windows, OSX, and Linux and simplification of the installation process.

Deterministic Versus Stochastic Cell Polarisation Through Wave-Pinning

Walther GR, Marée AFM, Edelstein-Keshet L, Grieneisen VA (2012): Deterministic Versus Stochastic Cell Polarisation Through Wave-Pinning, Bulletin of Mathematical Biology

I implemented Gillespie’s stochastic simulation algorithm in MATLAB and time-critical parts of the code in C. I further made extensive use of numerical bifurcation packages XPP/auto and MatCont.