CV: Andrew Cooke

Return home.


This CV is available (pdf) on the web — as a “general” CV it describes a wide range of projects. If you are interested in a particular skill, please contact me for more specific details.

Personal Details

Address: Europa 2189 Depto 22, Providencia, Santiago, Chile.
Phone (56) 6 038 070
Place of birth: Harrogate, Yorkshire, UK.
Nationality: British (permanent Chilean resident).
Languages: English; Spanish.


I have broad engineering experience and a strong background in physics. Several of my previous jobs have combined these — I have written numerical software in Java, C and Fortran — but I also enjoy technical challenges. I am very self–motivated, reliable, independent and productive. I also have many years’ experience in telecommuting.

I am particularly interested in domain–specific and `little’ languages / parsing / code–as–data / flexible configuration. These ideas often provide a good abstraction layer for building adaptable, maintainable systems. At a higher level, I try to involve the client in “lightweight” development — balancing iterative, adaptable development with fair estimates and open assessment of risks.

Languages: Java, Python, C, SQL (some OpenCL, Fortran, Clojure).
Platforms: Unix (Linux, (Open)Solaris), Windows 7, XP, 2000.
Web: Django, Spring, GAE, JSP, Javascript, YUI, jQuery, SVG.
J2EE, SOA: Mule, Spring, JBoss, EJBs (not entity), JMS.
Databases: PostgreSQL, Oracle, MySQL; JDBC, SQLAlchemy, Spring, IBatis.

Work Experience


Senior Software Engineer. ISTI, USA (Telecommute).

ISTI develop custom software for the geophysical research community; they are based in the USA but have engineers in several countries. On each project below I was responsible for design and implementation and usually interacted with the client (typically in parallel with one of the company founders).

Hardware / Numerical: A set of loosely coupled C programs that calibrate seismic detectors. These can be run separately, by hand, or under the control of a scheduler for automated calibration.}

GPU / Numerical: Optimisation of numerical Matlab/Octave code using OpenCL. Reduced calculation from 12m (Xeon CPU) to 10s (low cost NVidia GPU), shifting work from batch processing'' tointeractive data exploration’’.

Web / Database: Several projects constructing database representations of complex systems and then providing a variety of ways to access and manipulate that data — both directly (HTML, Ajax) and via additional services (REST, XMLRPC). Implemented with Java (Spring/JSP) and Python (Django/YUI).

One case called for an extensible system that could support large data volumes. The solution used Java’s Service Provider mechanism to support dynamic deployment of new data types, which were automatically mapped to the database. The schema was carefully designed to support application–level sharding.

Client Application: Python (WXWidgets) GUI to simplify management of remote data procesing system, including a ``map’’ of inteconnected components (auto–layout via simulated annealing).

Other: I have helped introduce a variety of ideas to the company, including the use of continuous integration, lightweight progress tracking, and wider test use. Most recently I launched — a static site generated with jekyll — as a way to share more of our work and raise the company profile.


Software Engineer. MuleSource, San Francisco, USA (Telecommute).

MuleSource is the company formed to support and develop Mule, an open source Enterprise Service Bus (ESB). I was part of a geographically–disperse team maintaining the core system, particularly TCP related transports.

XML Schema: I was also responsible for the main user–visible change in Mule 2.0: an XML-based configuration system using Spring’s extensible schema.


Scientific Programmer. CTIO, La Serena, Chile.

The team in La Serena was part of a larger development group based in Tucson, USA, that developed software for NOAO observatories.

SOA / ESB: The NOAO Science Archive was developed to store and retrieve astronomical data. I worked on analysis, design, implementation, testing and documentation of the system.}

This included assessing ESB systems and selecting Mule as a solution that provided good scalability, wide compatability with existing transports, and support for rapid development with Java-based messages — a good, future–proof balance for a SOA that was still largely internal.

I also developed a design approach that isolated business logic in POJOs. This separated messaging from the main code and allowed services to be `plugged together’ for simple, automated integration tests.

Numerical: I designed and implemented the Gemini/IRAF GNIRS Package, to process spectral data. This was based on the existing (but incomplete) NIRI package and implemented in IRAF CL/SPP (Fortran).


2002—2003 Head of development / Consultant. Webtron Finance, Santiago, Chile.

1998—2001 Software Engineer. Intertrader Ltd, Edinburgh / Leicester, UK (Telecommute).

1997—1998 Software Engineer. Concept Systems, Edinburgh, UK.


1995—1997 Postdoc. Institute for Astronomy, Edinburgh, UK.

1994 Research Assistant. CTIO, La Serena, Chile.

1988—1993 PhD in Astronomy. Institute of Astronomy, Cambridge, UK.

1985—1988 BA in Natural Sciences (Maths and Physics); Christ’s College, Cambridge, UK.

Personal Work

LEPL Parser: A recursive descent parser, written in Python, which uses co–routines to avoid blowing the stack on recursion. It also compiles simple ``leaf’’ matchers to regular expressions. Current (rather slow) development is based around extending the regular expression support to be non–exponential (see Russ Cox, 2007).

Dependency Arcs: In-browser SVG-based display of dependencies in a library of code.

User Icons: A service that ran on Google App Engine, to provide more attractive user icons. Died from a lack of market interest, but I learnt about GAE and the Data Store in the process.