Navigating the Next Frontier: An Enterprise Information Architect’s Ongoing Journey in Life Sciences

Written by Strategy & Solutions Leader, Life Sciences

Many of you may have read the LinkedIn posting announcing my new role at Fresh Gravity. I’m sure that more than a few readers are interested to learn why I accepted this role less than 6 months after announcing my retirement from AstraZeneca. I’ll use this article to share a snippet of my journey since retiring and what I hope to achieve in this new and exciting role.  

Firstly, after retirement, I travelled to Las Vegas, Hawaii, and New Zealand with my wife. This was a wonderful trip, but of course all good things come to an end, and I found myself in a cold, grey UK in the middle of December with wet and muddy footpaths that made it challenging to enjoy the countryside walks.  

Looking for something interesting to research, I scanned LinkedIn and technical resources. I became fascinated with the rapid pace of change relating to Large Language Models and was convinced that these introduced a truly significant change in Information Technology and perhaps society. Three things intrigued me at the time:  

1. What is the underlying mathematics underpinning LLMs? 
2. Why do LLMs hallucinate? 
3. Can LLMs, Knowledge Graphs, and Ontologies, supported by a strong backbone of master and reference data, be leveraged to improve the capabilities of an LLM? 

I found some excellent Google courses introducing LLMs and Transformers, but I wanted to understand more about the foundations. Lacking a formal background in Machine Learning and with plenty of time on my hands, I decided to invest some time in understanding the foundations.  

I found a wealth of recommendations for free learning materials on LinkedIn and progressed my learning of the foundations using the following resources.  

1. Linear Algebra (3Blue1Brown) – I have studied Linear Algebra before (>40 years ago), but the visualisations really bring the subject to life.  
2. Multivariate Calculus (Khan Academy) – Again, this is a topic I’ve studied and enjoyed before. Loved the material, though it was a bit stressful doing tests on this topic for the first time in 40 years. A key tip is that you need to practice calculus to avoid silly mistakes with minus signs.  
3. Statistics and Probability (StatQuest with Josh Starmer) – Another brilliant resource that also delves into Machine Learning and Large Language Models. I give it a triple BAM! 
4. Stanford Introduction to Machine Learning with Andrew Ng – I found this tremendously valuable as it emphasized the maths and statistics foundations of ML. Absolute highlight for me was solving the normal equation (X = (ATA)-1ATb) by hand. I’d never do it again (2 pages of matrix calculations are not to be taken lightly), but I learned a great deal! 

I skimmed several other resources related to Deep Learning and Transformers, but at that point my attention was drawn to some fascinating papers by Juan Sequeda and others, including Knowledge Graphs as a source of trust for LLM-powered enterprise question answering. This and similar papers have now become a new area for my research.  

What did I learn from all this?  

• Clearly, I need continued intellectual input as I move into my later years.  
• The capabilities of LLMs are absolutely astounding, but it’s not genuine intelligence; just applied mathematics and statistics. (In my view the capabilities are more to do with Linguistics and the remarkable properties of human language. Hallucinations are nothing unexpected, it’s just showing the limits of these techniques.)  
• My gut feeling about LLMs, Knowledge Graphs, and Ontologies supported with assured master and reference data seems to be true.

The latter brings me to the conclusion that the future of Enterprise Information Architecture with Life Sciences companies lies in seamlessly connecting assured master and reference data with ontologies and knowledge graphs, while leveraging them effectively in LLMs. 

Why Fresh Gravity?  

I have worked with Fresh Gravity and with Ajit Kumbhare in the past and have always found them to be innovative, focused and successful at implementing and integrating the range of technologies and capabilities that interest me. They also have a strong presence in the Life Sciences industry, which is my primary area of expertise and interest. 

Additionally, Fresh Gravity shares my view that the future for the Life Sciences industry should be to combine assured master and reference data, knowledge graphs, ontologies and Large Language Models (LLMs) to support scientific and enterprise questions. 

With that background, I jumped at Ajit’s offer to join Fresh Gravity as a Strategy and Solutions Leader for Life Sciences. Thank you for giving me this exciting new opportunity.  

For those who are concerned about my wife, this opportunity comes with sufficient flexibility to allow us to continue travelling and enjoying long-distance walks (we start the Southwest Coastal path in early May). 

The 3 areas I hope to develop in this new role are: 

1. Support Fresh Gravity’s clients by applying my 33 years of experience in the Life Science industry. I hope to achieve this by defining common patterns for strategic implementation and integration of assured master and reference data.  
2. Introduce new IT and data governance processes to support this vision of connected data. This may need new or upgraded tools and processes to support end-to-end use of data.  
3. Clearly define the role of an Enterprise Information Architect, a role that is poorly defined in most organisations, but in my opinion, is crucial for the future of the Life Sciences industry. 

Stay tuned for more blogs diving into each of these topics.