Providing data services consultancy for multiple clients.

As below, including:Multiple patent applications made, including some not yet lodged (details on request).Using Google Maps to identify building areas using image recognition techniques. Built interactive HTML/JavaScript front-end to allow selection of polygons including tracking of hierarchies.Managed project to use single view metrology to estimate the height of buildings based on a Street View image. Enhanced this offering by utilising a computer vision model retrained using transfer learning in order to detect windows which can then be used to count the number of floors in each building.Built asynchronous batch processing pipeline using Python-Celery, RabbitMQ and Fast API orchestrated using Kubernetes. This is monitored using Grafana dashboards with container logs stored in DataDog.Interviewed the entire organisation to gather requirements to create the organisation's data strategy.

As Lead Data Science role, but with line management responsibilities for one data scientist.Production workflow using Azure Storage Queues as trigger to run machine learning code inside Azure Functions allowing scalability at minimal cost.Managed a project to detect and remove solar data from mixed consumption/generation data using modelling techniques.

Wrote comprehensive data quality checker to parse and evaluate user uploaded datasets.Designed custom Multi Objective Optimiser to choose best building retrofit by minimisingenergy saved, cost and customer disruption using building data stored in MongoDB.Solved commercial building energy disaggregation problem using Scipy optimisation suite.Utilised Gaussian Mixture Models to carry out unsupervised clustering of building energy profiles and allow benchmarking with peer buildings.Used dilated CNN model to predict building energy usage from environmental data up to 3 months ahead, with a median CV RMSE of 17% and median NMBE of 3%.Built interactive floor plan visualiser in Bokeh, reducing time to quote from 2 days to 2 hours.Productionised all data science models using Docker, Postgres, FastAPI and Celery, allowing easy scalability from one to one hundred containers as demand increases.Acted as facilitator for KTP student.Three of the algorithms listed above were patented.Co-authored three academic papers.Manage third-party GDPR consultancy and ensure company remains GDPR compliant.Manage third-party IT supplier and all internal IT projects.

Promoted to Lead role once data science team was officially established.Data exploration and model creation of two engine failure prevention models built in R and productionised using Microsoft Azure, saving ~$300k in 4 months (still being used to date).Assisting third-party consultants and data engineer to build second version of the datascience platform using Azure Data Lake storage and Apache Spark.

Grew data science team from scratch to 6 members (interviewing and technical mentoring) in conjunction with a senior manager.Engaged with technical operations team to develop Aggreko's first data science model, to predict battery failures across 40,000 assets. Built in R and deployed via Microsoft Azure.Built the first version of the data science platform using R, SSIS, Azure SQL Server and AzureMachine Learning Studio, processing and cleansing on-prem data (MySQL) into the cloud.Completed various ad-hoc data science projects for a number of stakeholders (e.g. Head of Sales, Head of Product Development) including customer load profiling, top failure identification, alarm normalisation and job title vs conversion rate.Supervised KTP student from University of Strathclyde.Presentation of complex results to stakeholders up to CxO level using ggplot and D3.js.

Developed reports and dashboards in SSRS and PowerBI for existing BI solution across all areas of the business (finance, sales, inventory management).Developed SSAS models for new reports.Worked with Microsoft consultancy services for first proof of concept data science project.

Design and development of the company data warehouse (DW) in Microsoft SQL Server 2012; administering 170GB of data across 160 databases.Development of SSIS packages and stored procedures, including ETL processes for the DW and creation of data cubes from data stored in the DW.Creation of analysis front-end interface, including UserForms, to extract, manipulate and re-insert data from the DW using VBA for Excel.Automation of analysis and reporting methodology, reducing the time required to analyse each site on a monthly basis from 5 days to 1 day.

Produced a time domain model in VBA for Excel, with 95% correlation with turbine site data.Optimised a four parameter control system, increasing the system efficiency by 1.6 percentage points and extending the operating range by 30%.Produced data retrieval system using LabVIEW, Windows batch and FTP scripts.Automated data extraction from Citadel database using ActiveX Objects in VBA for Excel.Supervised one post-graduate and two undergraduate students.