In this episode of the Boss N' Data Podcast, Joshua Smith is joined by Andrea Rosales, PhD, Associate Director of Data Science at Blend.

Andrea has spent more than fifteen years across academia and industry. Her PhD at the University of St Andrews looked at how AI systems learn and adapt in messy, real-world conditions, and her work has moved from transfer and continual learning through to the reasoning and reliability of large language models. She is also co-founder of Mosaic, a Scotland-based community focused on AI governance, responsible AI and inclusive tech.

This is a conversation about the uncomfortable middle ground: not "AI will take your job" or "AI changes nothing", but what actually shifts when the modelling gets automated, the org chart changes shape, and nobody can quite say who signed off on the model.

You'll hear:

• The difference between robustness and reliability
• Why the data science bottleneck has moved from modelling to deployment, and what that does to how you build a team
• Andrea’s thoughts on what happens to junior roles when the repetitive work is automated
• Andrea's concern about critical thinking, from students who can't explain their own essays to developers who no longer own the reasoning behind their code
• What we're outsourcing besides intelligence
• Agent washing explained
• The agentic productivity paradox and how it plays out
• Three practical tests for meaningful AI accountability
• Data for good and the human-centric metrics that prove it

Disclaimer: The views expressed by the guest in this episode are her own and do not reflect those of her current or former employers.

Andrea Rosales, PhD, is Associate Director of Data Science at Blend, and an AI research and innovation lead with over fifteen years' experience delivering data and AI solutions across industry and academia.

Andrea holds a PhD in computer science from the University of St Andrews, where her research focused on how AI systems learn and adapt in complex real-world environments. Her work has evolved from domain adaptation and continual learning into research on large language model reasoning and reliability, with a strong emphasis on building AI systems that are explainable and robust.

She now works with organisations to design and deliver end-to-end AI solutions, with a focus on responsible AI adoption and data for good. She is co-founder of MosAIc, a Scotland-based community for people working on AI governance, responsible AI and inclusive tech.

👉 Connect with Andrea on LinkedIn: https://www.linkedin.com/in/andrea-rosales-phd

👉 Find out more about MosAIc Scotland: https://mosaic-community.uk/

Joshua Smith has been a Data and AI talent specialist since 2011. He is Sales Director at MBN Solutions, responsible for leading and supporting the growth and diversification of MBN's client portfolio, and building MBN's position as a leading Data & AI talent consultancy across the UK, Europe, and US.

Explore industry updates and the latest in data and AI talent on MBN Solutions' LinkedIn page.

Hosted on Acast. See acast.com/privacy for more information.

In this episode of the Boss N’ Data Podcast, Joshua Smith sits down with the MetaMetrics team ,Rebecca Shaw (Client Services Manager), Sam Watts (Director) and Philip Gaudoin (Director) for an honest conversation about the modern measurement paradox: marketers have more data than ever, and confidence in decisions has rarely been lower.

MetaMetrics is a specialist consultancy that has been doing econometric modelling (also known as marketing mix modelling) for nearly twenty years. They’ve seen the apparent death and the very real rebirth of econometrics, and they have strong views on why it has outlasted every shiny new thing thrown at it.

You’ll hear:

• Why the data problem changed but the underlying stats really hasn’t
• Analysis paralysis: when more data made marketers less confident, not more
• Measurement vs application: when the problem isn’t the model, it’s the adoption
• The honest take on where AI actually helps in marketing measurement and where it doesn’t
• Phil’s warning to stop outsourcing your thinking
• Vanity metrics, PPC and the new role of LLM impression data
• Brand vs performance: why econometrics is no silver bullet for brand and the client who lost 30p in every £1 to negative PR
• How to build real confidence with senior stakeholders

Disclaimer: The views expressed by the guests in this episode are their own and do not reflect those of their current or former employers.

Philip Gaudoin: Director, MetaMetrics

Phil joined MetaMetrics in 2016, bringing a commercial, consulting focus to the work. He started his career as an academic at the University of Cambridge teaching econometrics to undergraduates, before moving into marketing analytics in 2005 at Publicis Groupe. He’s been pulling threads between maths, marketing and management ever since.

👉 Connect with Phil on LinkedIn: https://uk.linkedin.com/in/philip-gaudoin-44714b44

Sam Watts: Director, MetaMetrics

Sam has worked in econometrics his entire career — first across media agencies, including Publicis Groupe, before joining MetaMetrics around 15 years ago. He leads delivery at MetaMetrics, with particular depth in the evaluation of digital and social channels and the pragmatic use of attribution modelling for clients across multiple markets.

👉 Connect with Sam on LinkedIn: https://www.linkedin.com/in/samuelawatts/

Rebecca Shaw: Client Services Manager, MetaMetrics

Rebecca came to econometrics from a marketing background, with a long-standing interest in marketing effectiveness and what truly drives performance. She supports MetaMetrics clients through the full lifecycle — making sure the analysis is digestible, the insights translate to action, and the relationship outlasts the project.

👉 Connect with Rebecca on LinkedIn: https://www.linkedin.com/in/rebecca-shaw-3752a711b/

Joshua Smith has been a Data and AI talent specialist since 2011. He is Sales Director at MBN Solutions, responsible for leading and supporting the growth and diversification of MBN’s client portfolio, and building MBN’s position as a leading Data & AI talent consultancy across the UK, Europe, and US.

Explore industry updates and the latest in data and AI talent on MBN Solutions’ LinkedIn page.

Hosted on Acast. See acast.com/privacy for more information.

In this episode of the Boss N’ Data Podcast, Joshua Smith is joined by Sean Harkin, PhD, Data Science & AI Lead at Lloyds Banking Group, for conversation about what it actually takes to move AI from proof-of-concept to production in a highly regulated environment.

Sean has 18 years across financial services from risk modelling in investment banks, through fintech, to leading data science and AI in the insurance arm of Lloyds Banking Group. He is currently shipping GenAI and agentic solutions into production where most firms get stuck.

You’ll hear:

• How Sean’s team controls LLM hallucination at production quality ontologies
• Why decision-recommendation tools in underwriting are designed to support humans, not replace them
• When to reach for a domain-specific model versus careful prompt engineering on a general LLM
• How to make AI reasoning transparent enough for regulators, risk teams, and reinsurers
• The cost-per-token reality check on agentic systems most teams underestimate
• What separates AI projects that ship from POCs that stall and the customer journey result one of Sean’s use cases demonstrates
• Why doing the science well and doing risk management well are the same thing

Disclaimer: The views expressed by the guest in this episode are their own and do not reflect those of their current or former employers.

Sean Harkin, PhD, is Data Science & AI Lead at Lloyds Banking Group, working in the insurance arm. He has 18 years’ experience in financial services across startups, banks, and consultancies, with a research science background spanning biochemistry, biophysics, and a PhD in applied statistical modelling focused on causal inference.

Sean now leads a team of data scientists, ML engineers, and other technical professionals putting GenAI and agentic solutions into production, cutting through the POC bottleneck that holds many firms back. His current work focuses on AI for medical documentation analysis, decision recommendation, and the practical realities of shipping AI safely in a regulated environment.

👉 Connect with Sean on LinkedIn: https://www.linkedin.com/in/sean-harkin-phd-22776674/

Joshua Smith has been a Data and AI talent specialist since 2011. He is Sales Director at MBN Solutions, responsible for leading and supporting the growth and diversification of MBN’s client portfolio, and building MBN’s position as a leading Data & AI talent consultancy across the UK, Europe, and US.

Explore industry updates and the latest in data and AI talent on MBN Solutions’ LinkedIn page.

Hosted on Acast. See acast.com/privacy for more information.