![]() |
. |
Altamar Networks - Contractor |
February 2003 – May 2003 Contracted Software Engineer [view the screenshots] Designed and developed a bespoke fibre optic link designer tool that allows users to painlessly model fibre optic links, analyse performance and output bills of materials. The tool is used to respond to customer bid contracts and also to be down-streamed to the customers for promoting sales. Therefore, stability and correctness were paramount throughout the design, implementation and documentation phases. The tool was developed as a standalone executable within Visual Basic 6, and included the following features:
To produce this tool, many hurdles were over come to solve such a complex problem space, which were mainly conquered by holding many transatlantic conference call and WebEx demonstrations. This project taught me a great deal about project management, RAD implementations and mainstream software engineering. Referee |
|
Martello Partners [Working for East of England Development Agency (EEDA)] |
April-2003
Financial Planner for the Demand Broadband (aka Connecting Communities Competition) Provided professional spreadsheeting and telecommunications to 8 of the 87 communities engaging in the Connecting Communities Competition within the East of England. This involved meeting the community representatives, discussing with them their dib response and providing the financial sections for the finished . EEDA is currently engaged in the judging process of some 87 completed bid reports to choose which communities will receive a portion of the £3 million government grant. Referee |
|
Altamar Networks |
April 2001 – September 2002 Senior Network Planner, Modeller & Customer Bid Response Worked for an American telecom ‘start-up’ company that designed and built electro-optical hardware for core communication networks. This job involved the detailed planning and costing of networks supplied by customers for bid response, along with supporting the engineering department with performance and cost comparison modelling. To accomplish this work I used a large number of software tools to route, optimally model and cost all elements of a network. I personally designed and built many of these tools myself using either VBA (interfacing with MS Excel) or C++, and ranged in complexity from simple Dykstra routings to complex simulated annealing simulation. The job also required a great deal of interaction with colleagues throughout the company including the CEO, CTO, sales & marketing to correctly plan and target our products. I was a key player in responding to over ten, multi-million dollar customer bids that often involved attending customer meetings worldwide. The amount of experience gained from this job role was truly immense. For demonstrational purposes here are three of many presentations slide packs (MS PowerPoint) avaliable for download:
Referees |
OPNET Tool This tool took a network topology and a set of demands, to calculate the best routes for each demand and calculates overall network configuration and dimensions. The output is then exported to the ‘Chooser’ tool for best technology placement and for equipment configuration and costing. The tool was able to take account of protection types, routing and grooming methods and contains simplified models of system and equipment configuration to enable it to optimally route customer demands. The development kit (ODK) using Microsoft Visual C++ was also used to extend the functionality of the tool. Chooser Tool [view the screenshots] I conceptualised, designed, implemented and evolved a tool called ‘Chooser’, which played a key roll in enabling Altamar Networks to respond customer bids. This tool took the completed routings and network topology from the ‘OPNET’ tool and interfaced with an Excel hardware costing spreadsheet to dimension, optimise and cost the network. The detail of modelling performed by this tool cannot be understated and flexibility for a wide degree of application. Chooser started off as a link-engineering tool that chose the cheapest or optimal hardware configuration of a fibre optic DWDM link. This utilised an Optical-SNR and DCF application algorithms to determine the best configurations for a number of transponder types and choose the most optimum. It was then further developed to optimise and cost whole networks with the inclusion of the OADM:placer and Vring Identifer algorithms Feature list:
|
|
BT (BTexaCT) |
September 1998 – March 2001 Senior Analysis, Unit Web Manager & IT Manager The main function of this job involved providing data & Internet traffic modelling/analysis to provide guidance in network dimensioning, service role-out plans and vender bid responses. This job also entailed:
Referee |
End-to-End Network Cost Modelling
This Microsoft Excel based model started out as a research project to see whether it was possible to create a generic network-costing tool that understood the requirements of a network from geographical point of view. The model development took the form of an evolutionary project that needed to be tamed to prevent to tool developing out of control. To generate realistic network costings, the usage of the given network is first calculated, which is provided by applying service usage to the residential and business population of the UK. The customer types used in this model are segmented into the known social groupings and sizes of companies, and stored with their geographical location within the UK. The hypothetical network architecture can then be calculated, starting with the access network and building up to the outer core, ensuring that the all customer requirements are met Fibre To The Home Model (FTTH) This model was commissioned to cost a fibre to the home deployment to all residentials within the UK, and was then further developed to cost other future technology deployments for cost comparison purposes. As with the End-to-End model this model is geographical based, as not two UK areas will cost the same, due to different customers densities and Infrastructure layout/deployment. Using a customer usage penetration of one a given technology the model executes iteratively through all the exchange areas within the UK calculating the costs of exchange piece of equipment and labour. To ensure that this is as true to real life, I created a geography builder that uses algorithms to calculate an access network architecture, outputting distance and customer usage information required for the costing process. BT Unit Website Within this job role I was the Webmaster of the Broadband Architectures & Optical Networks website held within the BT intranet. This has involved the creation of styles, GUI images, web pages and the administration of the entire site. The GUI of the website makes use of perl script to facilitate pass worded directories, and a Java menu that reduces the complexity and hierarchy of entire website. This website is hosted on a dedicated Linux server, which is my responsibility to maintain and develop. Currently the website is contains more than 1000 web pages and project documentation, and consumes about 2 gigabytes of storage space. With the recent break-up of BT, the website has undergone a style facelift to comply with BTexaCT design templates. |
|
BT [University Dissertation] |
November 1997 - March 1998 SmartQuill Project Programmer This job function entailed the monitoring of network demands and usage trends of BTs network infrastructure. This job used an in-depth understanding of the customer geo-demographics within the UK and how businesses of different sizes behave and utilise BTs data services, with a particular emphasis on the Internet I applied my software engineering skills to this work area to produce a number of robust modelling tools and utilities that are being used in anger within areas of BTs research and planning departments. About 50% of my time in this job was devoted to fire-fighting tasks that required rapid customer analysis of consumer and network data. All of this work was geographically based, as opposed to simple averaging mathematical approaches because it enables a high degree of accuracy and closely mimicked real world situations. Referee |
About the Smartquill Project
SmartQuill is a computer in a pen. It will have the capabilities of a PDA (Personal Digital Assistant) e.g. diary, data files etc but with the added advantage that it can recognise handwriting when written on ordinary paper without the need for an active surface. By dropping SmartQuill into its docking station, an associated inkwell, the data is transferred to a PC and translated into typed text. This allows working away from the office without the need to carry a laptop with you and when used in conjunction with a mobile phone can indeed become your mobile office. Sensors (accelerometers) in the pen converts handwriting into typed text through the classification of the electrical signal generated by the pen, similar to how speech recognition uses the output of a microphone. My Involvement in the Project My involvement took the form of an analysis, design and implementation of software code to convert the quill's movements into text, as my final year project for University. The analysis phase of the project looked into different methods of voice recognition and waveform correlation, with an assessment of their applicability for use on the Smartquill prototype. Many suitable methods were found to be fruitful, including Hidden Markov Models (HMM), which I deemed to be the optimal algorithm for implementation. The remaining development phases delved into the design of a prototype application which not only recognises written characters, but also allows the user to train the HMM algorithm to a users own handwriting. The results from my study helped prove that the movement to text recognition using accelerometers was feasible and that the BT project should continue. |
|
BT [University Year Placement] |
August 1996 - August 1997 Application Designer and Creator The industrial placement at BT entailed the creation of applications for demonstrating broadband platforms within a customer centre. The application design was such that the interfaces should be ‘fool proof’ and simple, so that users could obtain the real message behind the demo, rather than be distracted by complexity. Referee |
|
Company |
Dates
Job Function Job Details |