• Part-time consulting architect of music DSP project for music therapy. Responsible for prototyping, technical design and specification, developer recruitment, quality assessment and lead software development, leveraging multiple deep network models and ActiveMQ Artemis message brokerage for cloud based event driven architectures for audio signal analysis and processing.• Lead technical author of patent application, patent attorney liason and technical advice and review.(Python, Tensorflow, numpy/scikit-image, ActiveMQ Artemis, Docker, Wasabi S3, Jupyter, GitHub and GitLab CI/CD, Python poetry, VST plugins)

• Consulting to a number of clients in the audio, music and general software industries, providing practical experience in machine learning, DSP and software development. Clients not mentioned below include Emozia Inc., Muzieo Inc., Kenneth Willard Photography & Film, Aenvision Inc, and Bluestreak Holdings Inc.(Technologies listed with respective clients)

• Developed optimised linear AI models, and multichannel spectral and time domain feature extraction and DSP for embedded Analog Devices SHARC SC-58X processors for real-time SubPac tactile bass implementation of Jaguar/Land Rover “Body and Soul Seat” application, summarised on Wired, March 27, 2024. Development of training and test metrics and audio media dataset processing. Further real-time audio DSP development for Analog Devices SHARC SC-58X processors in C++ and OOP C.• Adaptation to real-time of audio “cocktail-fork” source separation algorithms.• Helm scripting of Google Cloud Kubernetes based GitLab CI/CD runners, for ADI embedded development systems, and MacOS/Linux/Windows CI/CD execution and testing of combined Python/C++/C codebases. MLOps helm configuration using MLFlow on GCP Kubernetes, and use for hyper-parameter tuning and optimisation of machine learning audio classification models.(Python, C++, Analog Devices CrossCore Embedded Studio, MacOS X, Xcode, Linux, Win10, MinGW, Google Cloud Vertex AI, PyTorch, Docker, Kubernetes, Helm, GitLab CI/CD, MLFlow)

• Member of a team developing a 3D video chat application using the Unity3D game engine with spatialised audio using Audiokinetic Wwise audio engine. MacOS porting of noise cancelling signal processing and Wwise plugins.(Unity3D, Wwise, C#, C++, Win10, bash, MacOS X, Xcode, VisualStudio, git, Concourse CI)

• Research and software development in machine learning based music production mastering.Research and development of audio sample recommendation systems, key, tempo, sample type, genre and instrument classifiers.(C++, Python, Boost, Juce, Numpy, scikit-learn, MacOS X, Xcode, MS VS, RH Linux EC2, git, Vagrant, Docker, bash, AWS Elastic MapReduce, DynamoDB, Elasticsearch, Hadoop, Hive, Spark, MXNet, Keras, Tensorflow, Sagemaker, DVC)

• Research and development of music information retrieval and machine learning for mobile music applications.• Developed Android audio synchronisation latency measurement and distributed data aggregation tools.(Android NDK, OpenSL ES API, C++, Java, Python, Amazon Web Services S3, elasticsearch, boto)

• Managed combined development and content team across three offices (SF, NYC, Berlin) using two week sprint based Agile/Scrum methods using JIRA, Confluence, Bitbucket tools. Responsible for product road map specification, developer recruitment, organisation structure.• Development of beat tracking and vocal transcription methods.• Implemented continuous deployment and testing architecture leveraging Vagrant and AWS CloudFormation.• Development of an AWS S3 based content management system for audio assets in Python using SQLAlchemy to enable music information retrieval algorithm operation and evaluation with a location transparent API.(Amazon Web Services EC2 Ubuntu Linux: Python, PostgreSQL, Tornado, boto, Essentia, vagrant, nginx, RabbitMQ, Redis)

Responsible for co-teaching the week long summer workshop in Music Information Retrieval at Stanford University's Center for Computer Research in Music and Acoustics. Lectured in content based feature extraction, beat tracking and tempo estimation, machine learning and information retrieval evaluation techniques. Preparation of iPython notebooks and Matlab code examples for labs and tutorials.https://ccrma.stanford.edu/workshops/music-information-retrieval-mirhttps://ccrma.stanford.edu/wiki/MIR_workshop_2014

Upgrade of audio signal processing routines for iOS guitar tutor app RockProdigy, performing real-time polyphonic pitch transcription. Implemented real-time audio compressor and reverb algorithms. Implementation of iTunes In-App Purchases, Facebook mobile SDK analytics. Code and data migration from legacy MS SQL server/ASP C# web app to LAMP stack.http://www.rockprodigy.com(iOS, MacOS X, Rackspace.com Ubuntu Linux: Objective C++, LAMP, Xcode, FlurrySDK, Facebook iOS SDK and ads API, cocos2d SDK)

With Imagine Research's acquisition by iZotope Inc., I became lead engineer of the MediaMined team at iZotope and then a member of the R&DSP team.• Lead developer of Agile/Scrum + Kanban team (using Jira/Greenhopper) bringing to market MediaMined Discover (http://mediamined.izotope.com/), a REST based, SOA web service for rhythmic and timbral similarity (content based search) of musical loop libraries. Utilized and implemented PCA, LDA, locality sensitive hashing and multiple distance algorithms. Built on AWS Elastic BeanStalk, EC2 with Cloudwatch process monitoring, RDS, SWF for cloud based signal processing control, SNS for automated testing.• Implemented cross platform C++, Python and shell script client side APIs for MediaMined, currently used in iZotope's BreakTweaker product.• Evaluation and outreach to academic community describing the MediaMined Listen audio classification system.• Designed and ran web based listener judgement task experiments to quantify A/B performance of similarity algorithms.(Ubuntu Linux, Amazon Web Services: Python, LAMP, Octave, Java, C++, Javascript, Boto, git, W3C XML Schema, XSLT, Jira)

We produce the Alphabet Soup sample player software and PocketFuel and PB&J sample libraries.(MacOS X, Win32, Linux: Objective-C/C/C++; Cocoa; Xcode; GNUstep; Altivec vector processing; Audio Unit API; Octave/Matlab; CVS, Subversion; MP3 codec)• Designed and developed; a low latency keyboard based sample and MP3 player/editor/signal processor consumer application.• Developed "Seeker440" -- an accurate realtime pitch detector Audio Unit plugin for instrument tuning.

Music Information Retrieval development on Imagine Research's MediaMined software platform.(MacOS X, Amazon Web Services, Ubuntu: C++, Java, Javascript, Python, Weka, Xcode, LAMP, svn, bash/awk)

Real time, low-latency audio signal processing for cross platform games.(MacOS X Xcode, Win7 Visual Studio, C++, Objective C, GNUstep, vector processing)

Working on the Quaero project in music information retrieval with a focus on analysis of musical rhythm.(Matlab, Weka, Common Lisp, C++, cmake)

Research and Software Development (Common Lisp, Objective C, C) in computational modelling of perception and cognition of musical rhythm.

(Win32, Linux, MacOS X: Objective-C/C/C++; GNUstep, Cocoa, USB API)• Cross platform, embedded systems application development of the "Thummer" USB expressive musical controller.

(MacOS X: Octave/Matlab; Objective-C/C; QuickTime Effects API)Developed “The Retail Experiment” — A real-time video installation at Diesel Denim Gallery, SoHo, New York. Luminance mask and faded captured still images into a prepared video stream using no external video processing hardware.

MacOS X, Windows 2000, Linux: Objective-C/C++/C;Smalltalk; Python; Cocoa/GNUstep/CoreAudio/CoreMIDI/PortAudio APIs; Project Builder/Interface Builder IDE; Win32: Visual C++, DirectMusic, DirectSound APIs; CVS); • Responsible for recruitment, training and management of five programmers. Responsible for liaison with technical and management staff at Apple Computer, M-Audio Inc. and open-source developers.• Designer and lead developer of "Ennio"​ -- a consumer application for automatically composing sound tracks to video, or video clips to music using automated analysis of video images extracting edits, motion and image tracking, and matching these against audio beats and phrases extracted from signal processing analysis of MP3 encoded music.• Lead developer for the open-source MusicKit/SndKit (http://www.musickit.org) — cross-platform music repre- sentation and synthesis project in Objective-C & Python.• Member of a development team writing “The Brain” — a music composition system. • Developed CoreMIDI drivers for M-Audio (http://www.m-audio.com) for their MIDISPORT USB MIDI inter-face devices.• System administration of HTTP, NFS filesystem, CVS version control, FAQ and QuickTime streaming serving using MacOS X-Server and RedHat Linux, SonicWall firewalling/VPN configuration.

(Solaris/SunOS, Linux, Windows 95/NT, MacOS: Microsoft C;SAS; Java; samba; amanda; sendmail; HTML); • Lectured in artificial neural network architectures and applications. Tutored in fundamental and advanced algorithms, concurrent programming techniques and analysis. Demonstrated in Java application and applet programming. Trained other system administrators on the administration of the Computer Science network.• Software maintenance and programming in Microsoft C for vision research in saccadic eye movements. Design and coding of an ECG wave measuring system in Microsoft C using a graphical digitiser and audio and visual feedback for rapid operator performance. Ad-hoc statistic queries and reports in the SAS statistical programming language for road accident research.

(SunOS 4: ANSI C)• Member of development team of a real time fiber optic cable monitoring and maintenance system.

MS-DOS: Microsoft ANSI C; Solaris: POSIX, C, BSD socket library);• Design and implementation of a custom aerial photography positioning system for the Western Australian state government department of lands adminstration (DOLA) interpolating shutter release times with GPS position. Conversion of graphics and peripherals drivers of a Geographic/Land Information System to device independent C graphics library.

Xenix, MS-DOS: Microsoft, Borland ANSI C; IAR, Introl C cross-compilers; 8051 and 68HC11 assembly language)• Designed and implemented a proprietary macro assembler and loader, RSA encryption algorithms, embedded system communications, hardware diagnostics and drivers for electronic funds transfer hardware and MS-DOS PCs.