Roche

Technical leadership across a federated organisation

At Roche, my work developed from engineering in a central enterprise data lake to technical delivery for Manufacturing Data Mesh. I served as Data Mesh Technical Delivery Lead while holding the role of Global Senior Data Engineer.

I helped shape the federated architecture across manufacturing sites in a regulated GxP environment. The work connected site-owned data products with shared platform capabilities, common engineering standards and clear operational responsibilities.

Data and ML for biopharmaceutical manufacturing

I engineered features from mixed manufacturing data: process and sensor telemetry from the PI historian, warehouse data and business context. The pipelines supported PyTorch and TensorFlow models of biopharmaceutical production processes, including titer and yield.

My work included telemetry downsampling, movement between systems and event-driven feature recomputation. Depending on the project, this used Pub/Sub, Snowflake streams and tasks, PySpark, Airflow and Cloud Functions.

GenAI architecture and reusable foundations

I built a reusable GenAI starter platform with Poetry, Cookiecutter, Kubernetes and AWS, bringing together CI/CD, MLOps and RAG orchestration. LangGraph and LangChain provided orchestration, with FAISS and Pinecone for retrieval.

I also authored a GenAI playbook covering architecture, implementation and evaluation, helping teams approach new applications with a shared technical foundation.

Leadership and delivery

• Defined SLOs and SLIs for AI and data products and contributed to architecture reviews with central platform teams.
• Led an engineering team on implementation quality and technical delivery.
• Developed a shared library of engineering components for ML systems.
• Mentored engineers through technical guidance and implementation reviews.

Python · PySpark · PyTorch · TensorFlow · PI historian · Pub/Sub · Snowflake · Airflow · Kubernetes · AWS · LangGraph · LangChain · FAISS · Pinecone