Incorporated to provide high quality consultancy services to clients in need of data science, AI, modelling or SAS expertise

• Created and managed a team of five data scientists to build, deploy and maintain a large catalogue of models• Implemented team standards for modelling projects, model review, code standards, infrastructure, deployment and third-party deliverables• Trained and deployed four neural network based PII redaction models for medical notes - Built a custom labelling interface and hired a team to label over 80,000 medical notes - Trained and tuned transformer and tok2vec models - Achieved 99.7% test recall and 99% precision - Created custom deployment to scale to redacting billions for medical notes• Deployed nine data science projects on three internal platforms with a focus on performance and reliability• Translated a legacy Neo4j project to Spark which reduced complexity by 30x and increased performance by 10x while improving reliability• Demonstrated a proof of concept for data monitoring that went on the be adopted by the Data Authority team for the entire data warehouse• Conducted retrospective cohort study of the relationship between pollution and pet health by applying academic best practice and using CMH and Cox PH models• Designed methodology for testing value of new pet healthcare product across 1,000 vets in the United States

• Scoped, designed and performed all data science work in a data driven project for a major publishing client• Created web scraping framework for multiple social platforms to collect hundreds of millions of connections• Implemented the most appropriate state of the art graph embedding techniques for network classification• Built latent class models to identify audience segments with distinct behaviours in R and Python• Produced price sensitivity model with appropriate visualisations to identify optimal pricing strategy• Produced and packaged custom model deployment

• Built prototype B2B product with multiple components including NLP embedding and ensemble models• Sourced, designed and implemented the ingestion of data to create a data library of over 300 features from public, private and commercial sources• Responsible for many aspects of a fast moving project from data science and data engineering to client updates• Drove the embedding of agile delivery principles in a young and expanding team• Supported and championed the development of junior data scientists and engineers as a strategic necessity

• Managed a small team of data scientists and analysts• Prioritised and scoped team workload with product owners and senior management• Developed the company's first live data driven model• Designed and delivered training plan for junior data scientists• Supported design of data science and data warehouse platform architectures• Model review and oversight• Agile team management using scrum

• Developed and deployed social network popularity prediction model to identify the optimal post contents- Convolutional neural network developed in Keras- NLP component using embedding layers- End to end build from design to deployment• Created forest based creative optimisation model• Automated hyperparameter and model selection for time series forecasting• Built models to provide predictive and prescriptive analytics offerings in the social media marketing space• Developed emotion classification deep learning model• Managed project to reconcile orthogonal data sources as a measure of data validity• Composed and delivered best practice training for data scientists using Google Cloud Platform• Built relationships across multiple divisions to foster data sharing and collation on GCP• Trained and supported junior employees in use of Python and cloud platforms

Data science and SAS development work with a focus of retail demand forecasting.• Independently built promotional sales forecasting models to predict the scale of uplift in daily demand for food products during periods of promotion - Identified and sourced key driver variables from a variety of data sources for use in ensemble models - Combined transactional metadata with sales and promotion data to build predictive features - Developed to fit the available infrastructure with data processing performed in Apache Impala and models in SAS and Python - Model candidates developed, tested and tuned in a custom model framework - Engaged with business stakeholders to design and specify the project requirements and goals - Guided deployment of model and creation of user review and promotion specification interface• Created new extreme weather effect model to reduce forecast inaccuracy and dependence on manual analysis• Developed tray size optimisation process to optimise logistics throughput without adversely impacting waste• Supported IT teams in the introduction of a new analytics platform through guidance and dataset design• Created reports to automatically identify where the largest business opportunities for tuning availability existed• Reviewed, improved and translated several models to realise the benefits of deployment on a Hadoop platform• Provided expert support and guidance to analytics team in a wide range of data and development areas including data processing, modelling techniques, statistics, SQL, Python and R• Provided insight into the relationship between delivery time and peak sales to identify opportunities for recapturing lost sales

The design and building of sales forecasting models using SAS tools including Enterprise Guide, Data Integration Studio and High Performance Forecasting, and connections to Teradata and Oracle databases.• Creation of an event sales forecasting system which features - Identification of events through cross correlation - Prediction of sales with bagged decision trees - Clustering of sales profiles with seasonal time-series models and hierarchical inheritance• The design and implementation of a sales forecasting framework with parallel routing through multiple model engines• Development of a modular testing application for measuring and comparing model accuracy through various metrics for use as an iterative model improvement tool• Investigation of data quality to identify discrepancies and reduce unnecessary sources of model error• Use of technical knowledge, implementation experience and a strong understanding of client requirements to help direct project design

The building and maintenance of treasury models to assess capital and liquidity requirements for internal and regulatory reporting. Responsible for all aspects of the model life cycle; from inception and development, through validation and implementation, to monitoring and review.• Created and delivered custom SAS training programme• Produced and maintained a library of SAS macros• Developed a range of Treasury models; interest rate and basis risks, swap rate simulation and others• Guided models to approval through strict validation and review processes

Implementation of liquidity management system. Development undertaken in Base SAS and SAS metadata tools including Data Integration Studio and Management Console.• Coordinated with project team to design, develop and launch the liquidity management system• Provided support to newly created liquidity reporting and management teams• Gained a firm understanding of the SAS metadata environment on GRID architecture• Recognised as a reliable provider of solutions to task automation and technical problems

Development of retail savings liquidity models as part of ILAA process. Defence of models under regulatory scrutiny. • Gained a firm and practical ability to program in the SAS language • Designed models to assess the liquidity requirements for a large retail savings book • Engaged with the FSA to present and defend models with favourable outcome