GORS Career Stories - Vlad
Hear directly from members what it's like being in GORS.
My background before joining GORS
I did a BSc in Security and Crime Science at UCL. This degree was a mixture of criminology, social research and mixed methods research, psychology and coding (statistics, machine learning, and with some optional modules I took web development and simulation too). There was also a work placement for 6 months where I worked part-time as a Research Intern at the Mayor’s Office for Policing and Crime in their Evaluation unit.
How I heard about GORS
I didn’t actually hear about GORS until I’d joined the Civil Service. But after hearing about Operational Research, that’s when it clicked that indirectly a lot of my degree had touched on OR principles, especially in evaluating crime prevention measures.
Why I joined GORS
A large part of my degree was focused on measuring outcomes after intervention, so I was already in the headspace of wanting to work on projects that lead to transformational real-world impacts. I am also analytically minded, and I liked coding during university so wanted to see how I could do it in practice in the real world. Retrospectively, OR was the perfect fit for me!
Separately I also had a mentor in MOJ which I met during university via a charity. He worked as a Project Manager so not GORS specifically, but he was instrumental in helping me understand what working in the civil service and MOJ was like.
My current role
I am currently an Associate Data Science Product Manager. I essentially look after product vision, ethics, managing stakeholders, identifying risks, and documentation for Data Science projects in Probation Data Science. Currently, I am focused on projects that use Large Language Models. It helps to have a technical background to understand caveats of models and how to communicate technical bits to non-technical audiences.
Before this product management role, I used to be a Senior Data Scientist here in the MOJ. I’ve previously worked on coding projects to do with large scale data linking across the criminal justice system, fines enforcement, reconviction rate estimation, extracting insights from employee data and prototyping a labour market dashboard. You learn something new in every new project and before you realise it, when you have a few projects under your belt, you’ll compare yourself to before and realise how much you’ve grown technically and in other skillsets like people skills, organisational skills and communication skills.
My proudest work moment
Two immediate projects come to mind.
Contact Log Semantic Search
The most recent one is my team’s launch of our Contact Log Semantic Search. Each year probation staff currently write millions of reports on offenders on probation (contact logs). These notes are unstructured, numerous and written in a variety of ways depending on the report writer. As such, it can be difficult for staff to search and find specific reports if they need to refer back to them (or even to hand over their caseload to new staff). Probation practitioners complete searches in order to find information about offenders, for example, to inform risk assessment or in preparation for engagement with the offender.
We trained a large language model on large amounts of text that are taken from multiple sources to calculate the strength of links between different words. Links can also be made between phrases or sentences. When the search tool is used, the model reads every contact in that person’s contact log looking for words or phrases which have a similar meaning to what was searched for. Each contact is scored by how strongly it is associated with the search query. A threshold has been set above which contacts are deemed relevant to the search and those contacts are returned.
The main benefit of the semantic search is increasing the amount of relevant information provided to staff when searching and a reduction in time taken to search by reducing the number of instances where users have to search a number of different terms to find what they were looking for (for example, arson, fire-setting, incendiary). By giving back more time to staff, they can spend more time engaging with people on probation and ensuring the best rehabilitation outcomes are attained, thus protecting the public.
Data linkage with Splink
The second project was helping enhance our person data linking efforts across the criminal justice system. In the criminal justice system, as with most other systems, administrative data is messy. For example, it’s not uncommon for there to be no unified identification number for an individual, or for an individual to get a new ID every time they interact with the criminal justice system. This can make it extremely difficult to estimate real numbers and statistics for purposes like estimating impact of interventions or their interactions with other government services like health, work and education, especially if the individual changes names (legally or with aliases), addresses and other personal identifiers.
Fortunately, our brilliant data linking team made the very popular open source package, Splink, which has over 10 million downloads globally. I was enhancing the outputs of an existing Splink person linkage by adding an additional data source.
Advice for prospective applicants
The first hard requirement for GORS is that at least 50% of your degree is numerate in nature, but aside from that if you’re curious that’s all you need, really. Next, it’s just about doing your curiosity justice and feeding it. I always encourage anyone who is curious about a role (in any profession or industry) to contact someone already doing it and seeing if they can spare some time to talk to you - you’ll learn much more that way! Who knows, you might also end up with a mentor.
Generally, I would say it doesn’t really matter what coding language you know (or don’t know) or previous experience. It’s more about your analytical aptitude – that’s what GORS is about. Can you think logically about the most appropriate analytical techniques and factors to consider? And after you’ve done your research, crucially can you practice your presentation and communication skills to present it in a digestible way (could be verbal, text, images)?
It’s all about practice and as long as you’re curious, disciplined and a little creative you’ll do well. For example, a lot of university students worry about not having internship experience. However, if they spent the 100 hours that they’d normally spend on internship applications on four 25-hour coding projects instead that would have been incredible on their CV, and they would have learned so much! There is so much open-source data, and you can also make synthetic data if needed.