This week’s Day in the Life comes from the burgeoning world of AI-powered marketing.
Neil Yager is Chief Scientist at Phrasee, a company that uses artificial intelligence and natural language processing to generate and optimise marketing copy.
Phrasee also happens to be one of the sponsors of Supercharged, a July 2017 event from Econsultancy which looks at exciting new AI technology in marketing. Do check it out.
Econsultancy: Neil, please describe your job.
Neil Yager: My role is at Phrasee is lead ‘data scientist’. This is a job that has only existed (at least with its own name) for a few years. A data scientist is someone who knows more statistics than a software engineer, but with more software experience than a statistician.
In my case, I’d add ‘research skills’ to the job spec. To me, research is systematically finding answers to problems that don’t have a known solution.
E: Whereabouts do you sit in your organisation?
NY: I’m one of Phrasee’s co-founders and work closely with the other founders CEO Parry Malm, our COO Victoria Peppiatt, along with our global team of developers, data scientists, and computational linguists.
Together we develop the present, and map out the future, of Phrasee’s technology.
E: What kind of skills do you need to be effective in your role?
NY: It is important to be ruthlessly analytical and data driven. I’m hesitant to take any action if I don’t feel there is enough evidence to support it. This applies to technical problems, but also to high-level business decisions.
At times this can make me a frustrating person to work with. However, Parry and Victoria are patient and somehow manage to put up with me.
E: Tell us about a typical working day…
NY: There isn’t really a typical day. I spend some days designing and tweaking machine learning models, sometimes I’m doing more traditional software development, and other days I spend reading academic papers to catch up on the latest developments in the field.
A lot of my time is spent thinking about stuff. It’s hard to explain what that tangibly is. In a previous role I was an inventor, and my job was to think of things no one had ever thought of before. Ever since then I’ve been pretty happy to stare into space and come up with new ideas. It’s these moments that really drive our innovation.
Image via Jean-Pierre Dalbéra - Le penseur de la Porte de l’Enfer.
E: What do you love about your job? What sucks?
NY: I love being at the cutting edge of technology. Some of the techniques we are using now didn’t even exist when we founded Phrasee a few years ago. This is a very exciting time for AI and especially for natural language generation.
On the down side, I work remotely (from Canada). Most of the time this works well since it allows me to bury myself in a problem and focus without interruption. However, there are times when I miss Phrasee’s legendary office banter and shenanigans.
Overall, I think being remote is a benefit. A lot of my job involves experimenting, analysing results and whatnot. So I’ll speak to HQ in London at 8am my time, and by the time they wake up the next morning, I’ll have had a full day to come up with ideas and solutions.
E: What kind of goals do you have? What are the most useful metrics and KPIs for measuring success?
NY: Ultimately, Phrasee’s success is our client’s success. Our goal is to help them to get a greater ROI from their marketing budget.
This is a double-edged sword. If our product is working, our clients immediately see an increase in revenue. However, if our product isn’t working, there is nowhere to hide. Therefore, my performance as Chief Scientist is tightly pegged to our customer’s results.
My personal goals are to use AI to do things people never thought possible. If you had asked me five years ago if AI could write better subject lines than humans, I’d have called you crazy! But here we are… and that’s what ultimately motivates me.
An intro to Phrasee
E: What are your favourite tools to help you get the job done?
NY: There is a programming language called Python that we use very heavily. We use this for natural language processing, server-side scripting, training AI neural networks, web frameworks, data visualisation, and much more.
Python is powerful, but is also a very graceful language that is easy to pick up. For anyone interested in dabbling in data science, I highly recommend doing some free online Python tutorials.
E: How did you get started in the digital industry?
NY: Prior to Phrasee I was working in computer vision, which was also the focus of my PhD research. I had no experience in the digital marketing area. Therefore, the story of how I got involved in the industry is the story of how I got involved with Phrasee.
To set the scene, AI and machine learning have been red hot for a few years now. Researchers with a strong background in these areas are in high demand and short supply. Therefore, I’m constantly approached by people with half-baked ideas for new startups. Normally, the pitch is along the lines of “Hey, I’ve got this great idea. It’s going to be huge! You can run with it and give me a cut.”
Phrasee’s CEO Parry is an old friend from university. His pitch was different. He said “I know this problem exists in the industry. If you can solve it, I can sell it.” He was unorthodox (to say the least), but he was driven, well-connected, and clearly brilliant.
When he introduced me to COO Victoria there was no doubt left in my mind. She has a remarkable ability to take a grand vision and make it a reality. I knew Parry and I alone would never get off the ground without Vic. The rest is history.
At first, I thought my lack of digital marketing knowledge was going to be a bad thing – but it’s turned out to be one of our best assets.
I don’t have any preconceived ideas about what’s good and what’s bad. So when my co-founders said, “We think X,” I could say, “Well, what about Y?” It’s this status quo challenging that’s allowed us to continuously innovate.
E: Do you have any advice for people who want to work in AI?
NY: From Phrasee’s perspective, AI is a set of tools that can be used to solve specific business challenges. This is a broad definition, and there are multiple entry points for those who are interested in the area.
To do AI research and development you need a very specific skill set, honed through both academic and industry pursuits. For example, I completed a PhD in in the area and have worked in AI commercially for many years.
This doesn’t exclude non-scientists though! AI companies are going concerns and there are many different ways to get involved. For example, language generation and understanding is a core research area of AI. Therefore, at Phrasee we have computational linguists who help develop this technology.
Also, we have sales people, customer success colleagues, and heck, we even have HR and an accountant. As far as I’m concerned, all of these people work in AI.
We are constantly hiring people who have AI skill sets, but also those who have other skill sets. AI, believe it or not, is only as good as the people driving it.
If you’re looking for a new position in marketing, advertising or ecommerce – why not check out the Econsultancy jobs board.