Harry became interested in machine learning (ML) while completing an astrophysics postdoc. Studying cosmology, he was regularly working with very large datasets, as he studied catalogues containing millions of galaxies. But Harry felt frustrated by the culture and work environment of academia. He decided to take a step back, asking what he enjoyed about his job as it stood.
”I loved working with data and delving into enormous lists of numbers that represented something in the real world, then trying to infer something about the real world from the patterns between those numbers.”
So for Harry, data science seemed a natural option. But switching from being a scientist in academia to a data scientist in industry was a big shift. Researching how to make this transition, he came across Faculty’s Fellowship Programme.
Products, deadlines and deployment: A new commercial mindset
Through Faculty’s Fellowship Programme, Harry started a placement in the data science team at Forestreet, an early-stage company based in London that has developed an AI-driven technology platform to research and analyse companies and markets. The team has a particular focus on Natural Language Processing. Harry’s six-week Fellowship project at Forestreet involves developing a deep learning model for topic modelling in order to improve the platform’s analysis of markets and companies. While the project is in a very different area to his work in academia, Harry has found significant overlap with many of the techniques and skills needed. “Before I was doing unsupervised clustering in some abstract space which describes, essentially, how well you could observe galaxies in the sky … whereas now, we are looking at how we can translate words into some mathematical representation and fill what we call semantic space, and then try to do some clustering in that space to figure out the relationships between words.”
But while many of the techniques Harry is using are similar, he has found the shift in mindset from academia to the commercial world very different. One of the main changes for Harry is the product-oriented nature of the work in industry, as well as the need to work to deadlines. Whereas in academia the models he worked on were rarely deployed, and the thresholds for completion were often arbitrary, “in the real-world you want to produce something that gets used quickly without trying to make it perfect.”
Small companies and big personalities
Harry also commented on the greater breadth of personality found in industry, compared with his experience of academia. “I prefer being in this environment where, for one, people don’t have that focus on politics and also are more diverse in the way they look at things.”
Something else Harry advised future fellows to think about was whether to take a placement in a bigger or smaller organisation. For Harry, the more personal and hands-on nature of a small early-stage company was something he was looking for and would recommend to potential fellows hoping to gain a breadth, as well as depth, of experience. “I had the opportunity to speak to directors and people in managerial positions who have years of experience.”
He also appreciated the chance to spend time talking with people across departments, including data engineers and the commercial team, an opportunity he would be less likely to have in large companies where operations are more siloed. Being part of a small and not hierarchical team, he also found more ways to run with new ideas and make significant contributions to the project, such as suggesting using the new Canberra distance metric. Forestreet was so impressed with Harry that the company offered him a full-time role as a Data Scientist during his Fellowship.
Advice for future fellows and hosts
Faculty’s Fellowship Programme is a great way to quickly make the switch from academia to data science. For Harry, the Fellowship’s support while making this transition, particularly the help preparing for and adjusting to the different mindset, was invaluable and accelerated his learning.
“In such a short time you can really learn the lay of the land. Now I can probably say I work on Natural Language Processing and not need to have my tongue in my cheek.”
Ben Holgate, Lead Data Scientist at Forestreet described the key benefits and the value that the Faculty Fellowship provides employers “Forestreet has been delighted to participate in Faculty.ai’s Fellowship programme to hire a high-calibre Data Scientist to contribute to improving our AI-driven company and market research platform. The Fellowship allows employers to benefit from Faculty.ai’s rigorous selection process that attracts top postgraduates and de-risks hiring through the Fellow working on a 6-week project designed by the employer. Moreover, Forestreet supports industry initiatives such as this that promote the training and growth of Data Scientists to boost the UK economy.”
Harry also highlighted the value of being surrounded by a supportive group of colleagues and peers. It can be an isolating experience working in academia, but on the Faculty Fellowship Programme, as well as at Forestreet, Harry relished the collaborative atmosphere, the constant back and forth and shared desire to learn. “Everyone’s asking questions all the time and everyone gets there quicker as a result.” He added that the fellowship lecture series were also invaluable, not just for the high-level of ML expertise taught, but because “they’re teaching you about the skills that you never picked up in academia, such as dealing with clients and working to deadlines.”
Harry would advise others looking to make the switch into data science “not to be afraid to ask questions and to get the most from the people around you.”
To learn more about Forestreet, click here.
