Samuel Atkin

Spearheading the creation of DSIT's first data science team, driving initiatives in digital policy, infrastructure, telecoms, data, and cyber security.

Cultivating extensive cross-functional networks to embed data science across the organisation and ensure alignment with business objectives.

Identifying strategic opportunities for data science to add value, including scoping projects and defining the team's direction and methodologies.

Establishing agile operational frameworks and championing a data-driven culture to support DSIT's technology and growth sectors.

Established myself as the face of data linking for the Better Outcomes of Linked Data programme, overseeing all data linking projects within the programme, and identify strategic opportunities for cross-governmental collaboration through data linking.

Led software development of the open-source product Splink, a tool for probabilistic data linking at scale optimised through distributed computing.

Led the development of novel internal data linking pipelines adding to MoJ’s cross-justice-system data using airflow, AWS glue, Python and Spark, joining together previously disparate datasets and enabling the creation of novel analytical products and functions.

Led analytical teams and expert statistical consultants to develop a set of multilevel models to model police resource demand and allocate £9bn of core grant funding to police forces.

Secured unprecedented buy-in from the entire policing sector to run a non-crime data collection from all police forces to allow innovative modelling of non-crime demand.

Led a multidisciplinary team of analysts, policy makers and consultants on the Police Activity Survey, surveying how all police staff and officers spend their time over a two-week period.

Created and developed a methodology for unit costings utilising police activity survey data to weight multilevel models and to ensure the demand of given crime and incident types is accounted for in budget allocations.

Wrote and presented monthly papers to the project’s Technical Reference Group and Senior Stakeholder group, whose members include police chiefs and their representatives, expert crime and statistics academics, and senior civil servants. Explained highly technical concepts to diverse audiences to bring them along the journey and create their buy-in to ensure the project's implementation.

Worked within an internal consultative function, to provide data science expertise and increase data science capability within Home Office analysis.

Utilised unsupervised machine learning to carry out geotemporal analysis into the relationship between key building locations and the occurrence of youth violent crime. Ran simulations of policing in these cluster locations to determine where best to place police/chaperones to project children.

Utilised natural language processing techniques to extract key information from a call for evidence around violence against women and girls.

Led the Home Office data science mentoring scheme to increase data science capability across analysts in the Home Office.

Led transition from physical databases to cloud data infrastructure, including the creation of a single unified premise-level database within Google Cloud Platform, utilising python and SQL scripting, as well as airflow to schedule updates of regular data releases.

Led a team alongside Ordnance Survey to create the Network Routing Model within PostGIS, to simulate supplier build and determine where to invest a £5bn government budget.

Lead analyst on delivery strategy for the UK Gigabit Connectivity Program.

Produced, deployed and maintained the dockerised R and Python applications to facilitate delivery professionals identifying and justifying government intervention within the telecoms market.

Produced DCMS’s 2018 Business Demographics Report during Covid-19.

Worked in Ruby and ReactJS to prototype novel software for smart meters from industry standard technical specifications to allow energy providers to interact with consumers, which has now been purchased by industry leaders.

Learnt CICD principles, test-driven development and deployment, version control, and the ability to utilise these skills to work alongside other developers on a single product.