Commercial Analytics at Scale in Pharma - From Hackathon to MVP

GSK are a science-led global healthcare company with a special purpose: to help people do more, feel better, live longer.

We have three global businesses that discover, develop and manufacture innovative pharmaceutical medicines, vaccines and consumer healthcare products.

In this talk i will share our experience in the Pharmaceutical business delivering commercial analytics going from hackathon to MVP.

From the initial ideas and business discussions through delivery of a hackathon as an accelerator, on to building an MVP. Using the Azure cloud platform and Databricks to rapidly ingest data and prototype.

I will touch on the challenges, opportunities and learning points of the process we went through to deliver commercial analytics at scale in Pharma.


1.WIFI SSID:Spark+AISummit | Password: UnifiedDataAnalytics

2.Commercial Analytics at Scale in Pharma: From Hackathon to MVP with Azure Databricks Peter Webb GlaxoSmithKline #UnifiedDataAnalytics #SparkAISummit

3.Introduction • GSK are a science-led global healthcare company. Our purpose is to help people do more, feel better, live longer • 3 global businesses – Pharmaceutical medicines, vaccines and consumer healthcare products. • Our goal is to be one of the world’s most innovative, best performing and trusted healthcare companies #UnifiedDataAnalytics #SparkAISummit 3

4.Initial Concept Cx board Hackathon idea conversation Senior leadership Business support from both challenge business & tech development 4

5.Hackathon prep Business Local business and Data prep - File per Demonstrate art of challenges - Clear tech teams engaged business challenge possible aims 5

6.Hackathon prep • Mixed teams – What – Business owners – How – Data scientists – Why – FLP (graduates) trained in Product concept • Strategic business partners 6

7. Hackathon Tech Azure Blob Azure Data Azure PowerBI Lake Databricks Data Sources File per challenge 7

8.Hackathon Event On the day Learning points • Mixed teams • Enthusiasm • Read backs at lunch and end of • More data…….. day • Prep teams more in advance • No Powerpoint • Senior leadership support & • Art of possible – attendance – Base working solution • Follow up agreed in advance – Business teams – Thunder Shiviah Databricks for his use of a Facebook Algorithm Out of 4 business challenges 2 accepted 8

9.Road to MVP • Focus on thread e2e • Business involvement with agreed aims • Three 2 week Sprints • Data ingestion – limited sets • …..yet still code with future in mind 9

10. MVP Tech Azure Data Azure Data Azure PowerBI Factory Lake Databricks Data Raw Sources Foundation Enriched 10

11.Scaling • Building on single e2e thread • Data ingestion – increase date range, volume, sets • Automation of pipeline • Model complexity • Outcomes focused backlog 11

12.Summary • Business engagement • Buy in at all levels • Ruthless focus on single e2e thread • Code discipline • Art of possible 12