MATLAB Expert

I have been using MATLAB since my days at university. Three of those years were working as part of the application engineering team at MathWorks. I've seen how MATLAB can be used from both sides of the fence and across a multitude of industries and application areas.

• 

High Performance Computing

Do more, or do the same in less time - the choice is yours.

• 

Effective Communication

I believe there is a place for simple and effective communication in any project.

Software Development

Broad experience with a variety of programming languages and environments. Including (but not limited to):

• C#
• Python
• Java
• C++
• JavaScript
• CUDA
• F#

Mathematical Background

My academic background is as a mathematician. I graduated from Oxford University in 2003 with a MSc in Mathematical Modelling and Scientific Computing. Modules studied included:

• Mathematical modelling
• Numerical optimisation
• Distributed computing for computational finance
• Numerical linear algebra

My dissertation explored the numerical solution of magnetic fluid flow and made heavy use of MATLAB and Fortran.

As an undergraduate I studied Mathematics at Cambridge University where focused on applied maths and theoretical physics.

• 

Finance Industry Experience

I have worked for a large asset management firm where I was responsible for taking quant research into production.

Prior to this I worked at MathWorks in a customer-facing engineering role. Working with a wide variety of customers. These included investment banks, hedge funds, and other financial services businesses.

• 

Ideas to Production

Go from novel idea to full implementation in days not months

Communicate in a Common Language

Widely used by quants and analyists for its power to explore ideas and quickly build up models. See how you avoid reimplementing in C++/Java/C#.

Keep core analytics and models as reusable MATLAB code. Deploy as standalone applications or integrate with other environments where appropriate.

Real World Data

In the real world data comes from multiple sources, each with its own format, is often incomplete and occasionally is plain wrong. Chances are your research and development code will fall over the moment it encounters an actual piece of data.

Industry Standard Development Practices

Apply proven productivity boosting techniques to the solutions your MATLAB users produce. This requires judgement and experience because you and your team want to keep the speed of development MATLAB enables without being slowed down by a traditional software development methodology.

We can select those practices which are most appropriate for the business needs and experience of your team. A few examples are described below.

Agile Development

Quick turnaround requires you to take on board modern software development practices. The key takeaway here is that the only way to deliver a successful project in a short timescale with uncertain requirements is to iterate.

Version Control

This goes without saying and yet it is still common to come across MATLAB code-bases which number in the 10s of thousands of lines which are not under version control.

Testing

We all know we should it more. They key is to be smart about it. Avoiding testing at the start of a project is a great way to buld up technical debt. Better still - automate as much as you can.

Continuous Integration

Take labour intensive tasks such as testing and deployment out of the hands of your developers and let a machine do it for you.

Documentation

Can be as simple as commented code to creating your own documentation integrated with the MATLAB help browser.

Turbocharge Your Applications

Do more, or do the same in less time - the choice is yours

Speed up MATLAB

Coaxing MATLAB applications into running as fast as they should is at its heart a simple step-by-step process:

• First up, ask yourself does it need to be any faster? Sometimes good enough is good enough
• Secondly, identify the bottlenecks. Never guess. Always measure!
• Now, apply your years of experience to optimise the existing code

Only then consider opting for the more exotic approaches such as:

• Take full advantage of your multicore machines
• Distributed Computing
• Graphics Processing Units (GPUs)

Not just for Games

Driven by the demands for increasingly realistic visuals the computational power of modern graphics cards continues to grow at a staggering pace. These massively parallel pieces of hardware represent a potentially cost-effective source of hardware for compute intensive tasks.

However, only certains type of problem are suitable for running on a GPU. And even those which are suitable can require a potentially prohibitive degree of developer effort and skill. A number of promising examples of middleware and tools are in development to make the process of developing for a GPU easier - this includes MATLAB with the Parallel Computing Toolbox.

I am available to help determine if a problem is appropriate and to help identify sensible technical approaches to developing a GPU solution. This includes developing low-level CUDA based solutions. Typically this would take the form of a proof-of-concept.

Develop End to End Software Solutions

Whether it's a compelling graphical end-user application, or a number crunching algorithm

Develop Complete Applications

Developing an applications should consist of so much more than delivering code.

Create Compelling Graphical User Interfaces

Developing a graphical user interface should never be an after thought.

Integrate MATLAB with External Code and Applications

Call out to existing code and applications. Whether you are using Java, C++, C#, Python, Scala, F# or anything else there is a way to make MATLAB communicate.

Go the other direction. Incorporate MATLAB code into your existing applications and infrastructure. It could be a critical algorithm you want to deploy to production or a whizzy calculation that you want your traders to be able to use in Excel.