Georghios Joseph

Lead data Scientist – 2022 October to present

Reporting to data directors and I provide data-driven services as capitalised assets. Managing a team, hands-on, focusing on innovation.

Senior Data Scientist – 2020 August to 2022 September

Reported to the head of product for analytics and data. Created data-driven products as capitalised assets. Managed a team, but also hands-on, focusing on product innovation.

Chief of Data Science – 2020 February to 2020 August

Built the company with C-level executives and technical staff. Devised business strategies. Acted as a communication bridge between technical and C-executives while managing expectations.

Chief of Data Science – 2020 January to 2020 August

Reported to the CEO and I helped built the company focusing on tech innovation. Board member for tech decisions made. Participated in meetings with commercial impact.

Senior Data Scientist – 2017 December to 2020 January

Streamlined and improved invoice analysis and enhanced staff training material from ML. Involved in devising the data strategy and culture, built, and led projects and teams conceived with C-executives.

Data Scientist – 2013 October to 2017 November

Project: data science informing chemical engineering for stem cell control in synthetic environments.

Data Analyst – 2012 October to 2013 September

Helped biomaterial/tissue engineering scientists develop therapies by understanding and designing surface chemistries of artificial environments to enhance cell performance.

 Collection Success – 2021 to 2022

Automatically re-collect direct debits intelligently using machine learning on transactional data from all transactional products. When enabled, it reduces re-collection cost by 5x, re-collects 94+% more DDs, and operates in seconds. The deliverable is a premium feature on the PCL platform that overrides the default re-collection date.

PCL Insights – 2020 to 2022

PCL as a platform is a black box for the customers. As our first data initiative, we return their aggregated usage, monthly, in a bespoke report available through the platform. We compared their historic performance along with an industry benchmark for adjusted for their company size and transactional volume.

Analytics platform built on AWS – 2020

Built with a team an easy-to-use analytical platform on AWS, utilizing tools like Lambda functions (Python), S3 for storage, and SageMaker for machine lerning. This platform was designed to be accessible even for novices, streamlining the data science workflow and enhancing the company’s service offerings.

Customer journey optimization – 2020

Leveraged POS data to identify bottlenecks, optimize store layouts and drive-throughs, and enhance preparation efficiency. Utilizing Python in Google Colab, BigQuery for big data, and TensorFlow for machine learning, Our suggestions had the potential to increase customer throughput by 30%.

VAT AI – 2018 to 2020

Implemented gradient-boosted machine classification for NLP tasks to identify invoice elements eligible for VAT recovery. This solution drastically reduced the time needed to process annual data from two weeks for one client to 30 minutes for 150 clients, saving millions for the NHS.

Forensics AI – 2019 to 2020

Utilized machine learning and subject matter expertise to detect duplicated invoices with slight mutations in codes, values, and dates, significantly reducing investigation time.

AI self-service platform – 2020

Developed a platform to house the above solutions with smart task queuing, enhancing the company’s digital self-serve capabilities.

Get-Chem AI – 2013 to 2017

GetChem uses machine learning to streamline the discovery of optimized biomaterial designs. It estimates cell behaviour from input variables from the surface’s chemical characteristics. The aim is to unravel the relationship between cells and biomaterial surface.

 Chem-ID AI – 2012 to 2013

Linking surface chemistry with cell reactions is challenging for designing of biomedical materials. The tool uses chemical data from the outermost surface of hundreds of synthetic materials. Key surface features affecting human stem cell adhesion can now be identified much faster compared to conventional methods.