💻 Mentoring Data Scientists, Goal Setting

We spoke to Conor McNeilly, Head of Data and Intelligence at GO!, to understand his process when it comes to managing and mentoring a team of data scientists, setting goals for data scientist career growth, and upskilling technical and soft skills. 

Quotations lightly edited for concision and readability 

🌐 From the Web 

UK IT Professionals express concerns about government’s generative AI goals 

A study conducted by O'Reilly, reveals that 93% of IT professionals in the UK are worried about the government's GenAI plans due to a lack of training and risk assessment. Despite significant investments in GenAI, workplace training and policies have not kept pace. 

The AI-amplified organization: Preparing for the large language model revolution 

The article emphasizes the need for businesses to invest in AI education and training to cope with forthcoming job displacements, increased productivity, and societal shifts, along with the importance of anticipating the economic and social implications, addressing ethical considerations, and promoting equitable growth in the AI-amplified future. 

How to Reskill Your Workforce in the Age of AI 

A conversation exploring the transformative potential of AI in various workplaces, addressing the need for a mix of technical and social skills in top management, and handling change management and employee fears. It explores the transformative potential of AI in various workplaces, addressing the need for a mix of technical and social skills in top management, and handling change management and employee fears. 

Could you talk about some challenges that come to mind related to making a transition into a management role for a data science team? 
 
I've been in management now for maybe six or seven years, [with] varying sizes of teams. In my last role, it was a team of seven and I came up through promotion. I started as a senior analyst. And then I managed to skip lead and go straight into manager. [I would say] the difficulty was around the transition from people being peers, [to] then line managing them, because we'd had that prior relationship where we were peers. I'd say we were still friendly, but the relationship dynamic changes. It's more formalized, [and] structured. 

So, finding the balance between that leading role whilst also maintaining a good open relationship and open communication was probably a bit of a challenge in my previous role. 

Also, mentoring. With bigger teams you can't just have one approach to everyone, cause everyone's unique, everyone's an individual [who] learns at a different pace. So tailoring [my approach]. I'm [personally] very methodical, quite process driven. I like things to be structured. Expecting people to do exactly that is not necessarily the right approach. 

So, finding those sweet spots of where they excel versus where they need a little bit of lift. I wouldn't say that it was a difficulty. That's actually the part that I enjoyed the most, but it certainly took time. I really love the people aspect of management, but it does take time to get into the right swing of things with people. 

On your point of tailoring the mentoring approach. It sounds like it might be time consuming. How do you manage that time to mentor each person individually? 
 
You are right it is time consuming, but I think putting that time in definitely pays dividends in the future, especially if you have a really good understanding of people's strengths and weaknesses and where they want to grow and how they want to develop. If people are happier, you have less turnover. The effort that you put in definitely pays off, the long-term quality of work gets better. So, I do think having a unique approach to everyone is important. 

The way I'm maintaining that time keeping is I'm not a micro manager. I never micromanage anyone else's time, but I [do] micromanage my own time. So, I schedule everything throughout every day. I have my checklist, so that I can then make sure that I give the appropriate amount of time to people.  

Some of the lead analysts might only need half an hour in a week whereas some of the more junior ones might want an hour. They might want two sessions, so it's just about putting that time in at the start, but then also having a glide path especially for the more junior people that OK, maybe you need an hour or two sessions this week, but in six months' time, I'd hope that you'd grow enough and that you'd feel more comfortable with me only doing one session or so. 

It isn't a constant time drain for want of a better word, but it's about having a good path where [when] people reach their targets, they become more in control of their own development overtime and with enough growth then you don't need to put in the same level of effort, you can point them in the right direction. 

Could you talk about that would look like, in terms of self-development resources available to team members? 

They become more of their own kind of information finders [but] it is also important to have the right platforms available to people. Especially when it comes to technical development, data science, analytics, so having the likes of data camp [or] other sources, training people to make good use of [platforms] like Stack Overflow even when it comes to their own technical development, it put less onus on me to do sessions with the team. 

When you're in a busy work environment, It can be a bit hard for everyone to find that time cause [we’ve] constantly got deadlines. but I would say that [it’s] something I always push in any team I've managed – to carve out time, even if it's just end of the day on Friday, even if it's only one hour. Just make sure you set aside that time to do something technical, whether it's with SQL or Python or Google Analytics, whatever they feel passionate about. 

You mentioned growth and making sure team members, say six months down the line, are able to jump off and figure things out on their own. How do you go about measuring that progress for an individual team member? 

I always like to put a clear plan in place. So, what is it that they need to achieve to be able to do their role well? What actions or outcomes can we use to measure that?  

It comes that they want themselves, so obviously, maybe some people want to learn more about Power BI and reporting tools, while other people might be more interested in stats and Python for instance. So it's not just about what the business needs, it's also about what people want for their personal development. Some people only care about maybe the tech side of things and don't really want to do anything in their future career plans with, [say,] people management. 

It's about very clear actions that you can take to show that you're growing, [for example] building your own power BI dashboard with or if someone was concerned with their attention to detail, then we can put QA steps in place, Integration testing, unit tests and things like that.
 
It's not just a kind of loose concept. There has to be something measurable against it that we can take and say yes, you have done that. There is a clear example of you doing this and this that to me shows that good progress.
 
There are softer skills that are harder to learn, [and] quantify progress, but I think it's just about continuously showing good examples and good working practices. 

Would you say there is a formal way of training these soft skills within your current organization? 

[Not] formal per se. It definitely goes into personal development plans. The way that I would like to approach it is leading by example. So showing what I mean when I say this is good storytelling or this is good kind of moderation of meetings or this is good a good way to scope out sessions or projects. 

So, I think for me I have to exemplify what good is and then over time give people opportunities to do it themselves, and I'd say I'd be around to step in if someone needs help or a little guidance. 

And then over time, you'd transition someone do that entirely by themselves when we know that they're confident, they're feel competent in what they're doing and they're getting good results. 

Earlier you mentioned helping team members figure out their career paths and their goals. What if there's a clash between the business need and the individual's career aspirations? How should one handle a situation like that? 

That is a tough one. There are definitely instances where, particularly [if a] business isn't growing massively, finding that opportunity, especially if someone wants to be a person manager, to move upward can be difficult. I do think it's you have to keep someone reassured that there will be a space, it might not be right now. We can look to anything that they need to do or learn or any kind of experience that they that might be missing in the meantime that way they can still grow, [but] they might not have the job title until it becomes available. 

But certainly, we can spend that time filling in any gaps, that would make them a better candidate, more likely to get that role when it does become available. So I guess for me it's still about focusing on the person's continuous improvement and then hopefully with them feeling like they grow, they're growing. But yeah, it is difficult. 

There [are generally] only so many leads or heads of department options that are available, and I think if you sit down with someone and really get to the crux of why they want to take the next step and what needs to happen before that, certainly we can help manage that growth and we can find additional opportunities elsewhere. 

It seems like a lot of it comes back to people making sure they're setting their expectations clear with their managers and both people working it out. 

Yeah. I'm doing that not just in the short term, but long term, so it's not just about what can I do over the next month. We need to think ahead. It has to be long term consistent planning. 

On a related note, do you find that data scientists have specific challenges when it comes to communicating with each other or communicating with stakeholders? Or is it just like any other role? 
 
I think it is just like any other role and I think again coming back to the individual person, some people are more comfortable talking in big crowds. I don't think job role has anything to do with that. 

You know, I've worked for some very technically minded people and yes, they don't necessarily want to become people managers. They're passionate about tech and learning and finding new ways to apply analytics or data science. And then you get some people who really do like the kind of personable aspect. I think that's completely separate to job role. It's just down to an individual. 

But again, it's that kind of thing that with experience and over time, even if you don't like it at the start, you know it's part of your role to do it because you need to be able to scope out a project, you need to understand what is expected, what the outcomes are, what success looks like and without good communication that's going to be missed. 

A final question. AI and LLMs have been in the news cycles quite a bit. Do you find that it's changing the way you go about doing things? 

The early stuff [perhaps not]. It certainly changing the way I want to go about business now. When it comes to trying to find efficiencies in day-to-day work, the likes of an AI assistant or Microsoft copilot. 

Being able to implement that technology, you're starting to see the efficiencies of it, it's going to save people time, especially [on] the creative side - altering or editing images, or writing, it definitely has a role in business that could help push efficiency. 

I think the early hype was like all our jobs are going to be replaced by a machine in five years. I think that's never going to be the case and there's a fine line, you know. I've been looking at certain AI platforms that do still need a human element to input the requirements and the needs and the outputs to check that it’s performing the way we want. So, yeah, big gaps that can make us work smarter, not harder. 

💻 Platform Highlight 

Data Literacy Project: A platform offering instructor-led and self-paced courses to develop the fundamental understanding of data. 

AI For Business Specialization Course: A course covering the fundamentals of AI and Machine Learning. Develop a deployment strategy for incorporating AI, ML, and Big Data into your organization. 

💼 AI in Business 

AI for Pricing Optimization – A Case Study 

This article on Talking Logistics explores the significance of price optimization and AI enablement in the context of Delly's, a major Brazilian food service distribution company. Delly's has a complex operation, serving over 200,000 customers with a vast product range and numerous distribution facilities. 

Key Points: 

• The Search for Profitable Growth: Companies are continually seeking ways to achieve profitable growth. Many have already invested in technologies like ERP, CRM, and supply chain management systems, but they are now looking for the next big opportunity to improve their business. 

• Price Optimization's Importance: Price optimization is identified as a crucial area for companies seeking significant business benefits. Setting the right prices can have a more substantial impact on profitability than increasing sales volume. 

• AI Enablement in Pricing: Delly's chose to focus on AI enablement in pricing due to its substantial impact on the company's bottom line. Pricing at Delly's is complex due to factors like commodity price volatility, a diverse customer base, and the involvement of category managers and sales teams. 

• Data-Driven Pricing: Delly's generates more than 800 million data points per month, making manual analysis impractical. AI is crucial for processing this vast amount of data to dynamically adjust prices. 

• Challenges with AI Enablement: The article highlights three major challenges in AI enablement: data orchestration, change management, and the need for extensive data analysis. 

• Benefits of AI Enablement: Delly's experienced significant benefits from AI enablement, including a 1.1 percentage point improvement in gross margin and the transition from state-specific to customer-specific price lists. 

This case study illustrates the practical application of AI in pricing optimization for a complex and dynamic business, resulting in substantial improvements in profitability and pricing strategies. 

💡 Word From Our Data Scientists 

Learn more about TeamEpic’s Data Talent here. 

🤖 Prompt of the week 

Act as a natural language processing expert. I have a text dataset [describe dataset]. Please help me build a text classification model using BERT. 

Ex. Description: This dataset contains customer reviews and ratings of various smartphone models. It's a collection of text data where customers have provided their opinions, experiences, and feedback about smartphones they have purchased. The reviews often include both text and numerical ratings (e.g., on a scale of 1 to 5). 

See you next week,

Mukundan.

Do you have a unique perspective on developing and managing data science and AI talent? We want to hear from you! Reach out to us by replying to this email. 

📊 Community Poll

We’ll share the results from this poll with you next week!