Maths, museums and building bicycles: Harrison Pim
Who is Harrison? Tell me a bit about yourself…
My work revolves around data science and machine learning these days, but it started with a degree in physics. During my master’s I worked on a set of quantum chemistry simulations, and, almost by accident, fell in love with the maths which underpins most of the machine learning (ML) advances over the last few years.
Work in ML felt much more enticing than where I was heading in academia. I still wanted to spend my time on weird, science-based problems after finishing my master’s, but academia felt like a poor fit for me culturally, and data science offered a more immediate way to make a difference in the world. I ended up working on digital teams in museums, and most recently at Wellcome Collection, where I helped to build their search engine and knowledge graph. That work introduced me to both modern and historical approaches to information retrieval, which I also found fascinating.
Can you tell me a bit about what led you to Climate Policy Radar?
I’d always been interested in working on climate. The problems are huge, they’re difficult, and they require large-scale collaboration, all of which are intellectually exciting.
I found out about the role through Kalyan (CPR’s Principal Data Scientist), who I met when we were both working on ML projects in the museum sector. We had been following each other's work for a couple of years and when a job at Climate Policy Radar came up, I was already familiar with their goals and ambitions. I’d seen their prototypes and was impressed by the open experimentation happening around search technologies. CPR’s ‘Labs’ ecosystem (https://labs.climatepolicyradar.org/) and their work around the Global Stocktake (gst1.org) made me realise how seriously they were tackling the challenges in finding and understanding climate policy.
What excites you about Climate Policy Radar?
I was initially drawn to CPR’s mission, and particularly the radical approaches to information discovery and retrieval that they were experimenting with. Being able to contribute to such an ambitious goal, particularly in the climate space, really appealed to me. The fact that CPR takes the same radical, experimental approach to its working culture was also appealing – I’m a big fan of the 4-day week!
Since joining, the people have been a huge part of what I like about this job: coming to work each day with such a kind and intelligent group of people is a joy.
What does a day in the life of a data scientist look like?
My work varies a lot from day to day. It’s all quite collaborative: I spend a lot of time with our policy team, helping to structure their climate expertise in a way that our machine learning models can understand. I also work with our team of data engineers, who make sure that the models we’re developing will work on all of the thousands of documents that we look after, and with our product managers and designers who shape the way that users will interact with our data.
In our more independent, research-driven work, we’re often trying methods for the first time, so the job inherently involves a lot of unknowns and uncertainties. It often feels a bit like I’m trying to build a bicycle from scratch, in the dark, without a manual. I’m usually trying to create a tool for other people to use, but essential components often need to be custom-built, or reshaped to work within the rest of the system. I try to remind myself that success isn't just about building something functional: if I’ve done my job well, the tools we build should also feel fun and intuitive for the people who use them!
Where can you be found on a Friday?
I live in Brighton and I’d love to say that I’m often found down on the beach, or out for a run around the South Downs. The truth is that on a Friday I’m probably writing code for silly personal projects, or watching films with my cat, Sprout.