global optima

Working on new Totara based projects
PHP, Postgres, Totara & SCORM

Development of a heavily modified Moodle with 100k+ users integrated with SAP
PHP, MySQL, Moodle & SCORM

Development on a number of Moodle sites, concentrating on custom reporting
PHP, MySQL, Moodle & SCORM

Development of a Facebook app in Flash using Amazon web services (S3, Cloudfront etc)
FBML, FBJS, Flash, AS3, AWS

Porting v5 modules to v6 (integration with ESiT database)

Drupal 5 & 6, ESiT

Enhancing & testing 2 Symphony CMS projects for Random House Publishing
PHP, XSLT, MySQL + CSV/XML processing

Co-development of a number of Moodle sites
PHP, MySQL

Technical lead and sole developer on www.raggs.tv
PHP, Flash 9, Actionscript 3 + Custom lip-syncing engine + CDN

Prototyped database for storage of case data
PHP, MySQL, ExtJS, Google Maps, Tagging

Note: Due to illness I was unable to see this project through to completion

Development of internal training portal for Bupa (custom e-learning)
PHP, jQuery + third party proprietary portal ramework

Development of IRS exam engine + admin site
PHP, jQuery, MySQL + ongoing support

Research & development
Built boutique for selling personally compiled digital products

Co-developed www.suzuki4.co.uk
Flash 8 framework + site editor for all depos

Extended & bug fixed www.samaritans-purse.org.uk
PHP, MySQL, Drupal

Flash game for Andrex
Highscore board, 2 levels, psuedo-isometric environment with 2 ai opponents

Development of 'Kratts Creatures'
Flash 8, Actionscript 2 + Bespoke video sequencer & sharing

3 core sites (IPE, HSS, HSM)
3 journal sites (AIR-USER, HSD, SHE)
1 exhibition site (HSE)

built with GO:Independence
PHP, MySQL (w/ replication), PEAR + integration with BusiPro

2 core sites (WEE, CM)
3 journal sites (EES, WES, PCS)
1 exhibition site (EE)

built with GO:Independence
PHP, MySQL (w/ replication), PEAR + integration with BusiPro

Conversion of Delphi application to classic ASP
ASP, HTML, CSS

Non-flash alternative games for Teletubbies
Javascript

Java client-server application for creating online slideshows

Co-founded with Poisson Rouge SARL & SnEm Limited
Technical director

Co-founded with Poisson Rouge SARL
Umbrella collective for Global Optima & Poisson Rouge projects

Co-founded with Paul Winterhalder (now of PR23 Limited)
Creative director
Closed up in 2005, remaining assets & clients migrated to Global Optima

A personal project that utilises jQuery, PHP, MySQL, Python, OpenCV and some custom jiggery pokery.

Basic premise is a simple image 'work shop' based in the brower where I can render different segmentations of images. Segmentations are generated on the fly or through a background process,

Features are detected via Python with both FAST and Prewitt corner detection, face dection also supported through OpenCV haar cascades (but not yet added).

Feature rendering with connected nodes is all done with HTML 5.0 canvas element.

Also allows for tagging with automated Exif extraction. Tags are key:value pairs split into groups, keys support hierarchy schemes like Location:City, Location:GPS or Location:GPS:Longitude etc.

Tags can be spread from a single image to a whole gallery, also supports duplicate tags (e.g. a Name tag can be applied multiple times for different people in an image).

Next few jobs:

• fleshing out tagging ui
• speed up image processing (multi-core or cuda)
• add other interesting segmentation algorithms
• add face recognition segmentation
• allow manual association of features with tags (i.e. faces, groups of nodes etc.)
• extend network rendering tool for more creative skeletal structures

• 

R&D for a potential project.
This was my main contribution for the proposal, an ant game
Process involved creating additional tools to define 'pheremone' paths for letters and shapes

Currently in the process of reworking and modifying this code for other purposes

• 
• 
• 
• 

• 
• 

• 
• 
• 

• 
• 

Simple strobing lights

• 

Simple circling spots

• 

Simple flv buffer visualiser
Note: Video is of Hendrix impersonator who was hovering around Brighton, England

• 

simple yet aesthetically pleasing

• 

watch the polys contract as they touch each other

• 

responding to simple touch sensors

• 

use of sin, cos and tan to generate points in a 3D space, then connected using a slightly offset line

• 

a simple demonstration of how Processing can be used to create interesting user interfaces.

• 

an unpublished paper I wrote whilst at HC2 on path finding

• 

One of my current interests is the management of tags on photo sets. Usually a set will be between 20 and 200 images, thumbed onto a 1920x1080 wide screen monitor at about 100x100 pixels with vertical scrolling. Thumbs are selectable, deslectable, CTRL + A for all and drag select enabled with a couple of jQuery UI libraries.

In this project, tags are associated with an entity, there are 3 simple entities which represent the file structure:

• Album
• Set
• Image

Each entity is a db table with at least an id and name column. Tags in this system are Key:Value pairs, there is scope for meta data per tag possibly to be implemented at a later date.

A system table 'sys_tags' contains a list of all tags in the system across all entities:

• tname
• calias
• cname
• cgroup
• ctype

Tags are essentially table columns with this sys_tags table keeping track of everything, since all 'tags' must comprise at least a key and value, we can map the key to a column name and the value as a record value. Bare in mind that if I create 5 tags A,B,C,D and E against the images entity, each image can take on a value for each tag, of course this creates a massively sparse table which I'm accepting of for the time being.

Note: Why accept such a sparse table structure? Right now I'm interested in a tool for managing tagged data, I've tried a few different approaches and this is just a new model. Please read more about tools for this model before passing any judgements.

So a case in point where we pull EXIF meta data from each image and map it to a tag, different camera manufacturers add very different data and even similar data can be tagged with different keys and values can be represented in varying foramts. A straight copy+paste of the EXIF data from a set of about 2k images from different cameras resulted in about 200 columns in the images table, a snapshot of 10 tags/columns:

19 	exif_make
20 	exif_model
21 	exif_orientation
22 	exif_xresolution
23 	exif_yresolution
24 	exif_resolutionunit
25 	exif_software
26 	exif_datetime
27 	exif_ycbcrpositioning
28 	exif_exififdpointer
29 	exif_thumbnailcompression
30 	exif_thumbnailxresolution
31 	exif_thumbnailyresolution

You can see all these tags are prefixed 'exif_', this matches their group which allows me to seperate user-created tags from automated tagging.

spreading tags current selection current set