ОБ ЭТОМ ЭПИЗОДЕ
👉 Discover how Cambridge Spark helps organisations build the data and AI capabilities needed to turn strategy into measurable impact: cambridgespark.com
In this episode of Inside the Algorithm, Dr Jeremy Bradley sits down with Dr Gueorgui Mihaylov, Principal Data Scientist at Haleon and Visiting Research Fellow at King's College London, to explore what happens when rigorous mathematics meets the messy reality of a global consumer health supply chain.
Gueorgui walks through two of the most technically ambitious AI projects running inside Haleon today. First, the AI Inventory Planner, an ensemble of machine learning models, survival analysis, and stochastic simulation designed to reduce inventory holdings by double digits while maintaining or improving service levels. Second, the Golden Batch project, an attempt to bring real-time process control to pharmaceutical manufacturing using data-driven corridor modelling of a highly non-stationary, non-Newtonian fluid system.
The conversation covers global forecasting algorithms, Weibull-based delivery modelling, principal component analysis for process visualisation, and the long road from working prototype to enterprise adoption.
Follow Data & AI Mastery so you never miss a future episode of Inside The Algorithm.
If you want to learn more about how AI is revolutionising demand forecasting and supply chain optimisation, listen to the Data & AI Mastery episode between host Dr Raoul Gabriel Urma and Peter Laflin, Director of Data and Analytics at Morrisons:
Spotify: https://open.spotify.com/episode/6HcwcdIGOguBZp1vADE1Z0
YouTube: https://www.youtube.com/watch?v=NKdjiYVC9H8
Glossary Terms:
Inventory Management: The systematic process of ordering, storing, tracking, and controlling a company’s goods to ensure the right items are available in the right quantity, place, and time.
Stochastic Phenomena: Events, processes, or systems that are inherently random, unpredictable, or probabilistic in nature, rather than deterministic.
Demand Forecasting: The process of predicting future customer demand for products or services using historical sales data, market trends, and analytical methods.
Out-of-sample bootstrapping: A resampling technique used to estimate a model's performance on unseen data by generating multiple datasets through sampling with replacement from the original data.
Non-Newtonian Fluid: A substance whose viscosity changes when subjected to stress, agitation, or shear force, rather than remaining constant.
IoT Device: A physical object embedded with technology, software, and sensors that allow it to connect, collect, and exchange data over the internet or networks without human intervention.
Chapter Markers
(00:00) - Episode Introduction and opening insight from Dr Gueorgui Mihaylov
(04:37) - Setting the scene: What is the AI Inventory Planner solving?
(07:17) - The challenge of supply chain unpredictability and stochastic buffering
(12:26) - Scenario simulation engine and constraint optimisation
(17:00) - Scaling the tool across a global enterprise
(19:17) - Introducing the Golden Batch project in pharmaceutical manufacturing
(25:00) - System architecture: data historian, execution environment and visualisation
(30:57) - From monitoring to closed-loop control: stages of adoption
(34:28) - Open problems at the frontier: geometric deep learning and LLM orchestration
(37:02) - Advice for practitioners entering industrial AI
Useful Links
Connect with Dr Gueorgui Mihaylov on LinkedIn: https://www.linkedin.com/in/gueorgui-mihaylov-891278197/
For more AI insights follow Jeremy on LinkedIn: https://uk.linkedin.com/in/jeremy-bradley
Explore Cambridge Spark’s AI upskilling programmes at https://www.cambridgespark.com
