victor 🌙 contact me 📬
data scientist
Since September 2024, I have been working as a Data Scientist at the MERCK APAC Healthcare Regional hub, based in Singapore. In this role, I focus on driving and leading innovative AI projects to enhance strategic decision-making. And unlock actionable insights across pharmaceutical markets in the Asia-Pacific region.
During my mission, I mainly collaborate with the APAC Insights & Analytics Director. We think, create and deploy AI-driven initiatives in order to Leverage underutilised data. This position allows me to work on a wide range of projects, from data analysis to AI development. I'm also responsible for the technical implementation of the projects. My superior comes up with a problematic and I come up with a solution leveraging internal tools.
  • Created a new source of regional market insights by building a RAG-based LLM app to extract key points from unstructured pharma market reports (Python, Streamlit, OpenAI, Qdrant)
  • Leveraged LLMs to translate and summarize field-force feedbacks (+300/month), surfacing key qualitative insights and improving cross-team alignment around multiple countries (Snowflake, LangChain, Claude)
  • Automated personalized country and regional level market analysis, accelerating strategic decision-making for teams around the world (Python, OpenAI)
details
These projects were part of a broader initiative to embed AI tools into the regional analytics workflow. Created a new source of regional market insights by building a RAG-based LLM application that extracts key points from unstructured pharmaceutical market research reports (~70 pages each). Previously a manual, time-consuming exercise across 11 APAC countries, the automation enabled faster quarterly insights extraction and streamlined insights sharing, freeing up shareholders for higher-value strategic work. The result is shared through a centralized Power BI dashboard. Allowing teams to track and analyse the evolution of their country insights over time.
Leveraged LLMs to translate and summarize field-force feedbacks (+300/month) collected across multiple countries and languages. The initial data was fragmented and difficult to interpret, limiting its strategic value. Using Snowflake, LangChain, and Claude, the pipeline automated translation, regrouped feedback by theme, and extracted key qualitative insights that were shared via a centralized dashboard, improving clarity and cross-team alignment. This initiative also encouraged countries to improve the consistency of their field-force feedback collection and helped spread this habit as new countries joined the program. (Snowflake, LangChain, Claude)
Automated personalized country and regional level market analysis, enabling stakeholders to quickly assess key performance metrics for both Merck products and competitors across flexible timeframes and indicators. By streamlining access to consistent insights, the solution accelerated strategic decision-making and supported better feedback discussions across markets and global teams. The project gained visibility at the global level, and other regions are now requesting to scale it to their markets. (Python, OpenAI)