Preconference Workshop
This hands-on session is designed for ML enthusiasts and biostatistics students who are familiar with the theory of ML but new to the world of operationalising and production deployment. Spanning healthcare and life sciences use cases, this workshop bridges the gap between experimentation and real-world impact.
Participants will learn the full ML lifecycle—from model development to deployment, monitoring, and feedback—through relatable examples and simplified end-to-end workflows. The session includes:
- Walkthroughs of lightweight production pipelines
- Introductions to essential tools (FastAPI, Docker, MLflow)
- Group exercises to help learners think like ML engineers
By the end, attendees will understand how to move from model building to creating scalable, maintainable, and production-ready ML systems that work in real-world conditions.
- Healthcare, BioStats students, and early-career researchers
- Data scientists and ML practitioners transitioning from notebooks to deployment
- Anyone experienced in ML modeling wanting to understand the operational side
Sushant Penshanwar : Sushant Penshanwar is a Senior Data Scientist at ReliaQuest, a leading cybersecurity company, where they focus on applying AI to enhance threat detection and decision automation. With nearly a decade of experience across global organizations like General Motors, Pfizer, and Cipla, their expertise spans Natural Language Processing, Classical Machine Learning, Generative AI, and Agentic AI.
During their tenure at Pfizer, Sushant led impactful research on biomarker selection for multiple myeloma using machine learning — a project that deepened their commitment to applying AI in high-stakes, real-world domains. Currently, they are focused on the practical challenges of taking machine learning models from experimentation to scalable, production-ready systems.
Sushant will be speaking on the practical frameworks and challenges of productionizing ML, drawing from their diverse cross-industry experience.
Every clinical programming team is being asked the same question right now: What does AI actually mean for this work? Most answers fall into one of two extremes: AI as a replacement for the programmer, or AI as a tool with no place near submission-grade work. Neither survives contact with a real ADaM dataset.
This workshop demonstrates a third perspective through live examples. It shows how a large language model working alone can generate code that appears correct yet misses critical nuances that experienced programmers routinely catch.
Attendees will watch the same ADaM derivations and TFL outputs built twice: first using AI alone, and then using AI guided by a human-authored framework of standards, conventions, and review principles. The contrast highlights how human expertise transforms AI from an automation tool into a productivity amplifier.
Rather than promoting or dismissing AI, this workshop provides a practical and balanced demonstration of Human-in-the-Control Programming, showing how experienced professionals can leverage AI while maintaining quality, regulatory compliance, and scientific rigor.
Participants leave with a repeatable framework they can apply immediately on their own studies and projects.
- How AI-generated code succeeds and where it commonly fails
- Why human judgment remains essential in clinical programming
- How to build effective AI scaffolds and programming conventions
- Techniques for reviewing AI-generated ADaM derivations and TFL outputs
- How to implement Human-in-the-Control programming workflows
- Practical approaches for responsibly integrating AI into clinical development
Avinash Bandi
Senior Advisor – Strategic Capabilities & Innovation
Eli Lilly & Company
Avinash is a Mechanical Engineer by training with nearly two decades of experience in clinical statistical programming. His expertise spans ADaM, SDTM, and TFL development, with a strong foundation in SAS and advanced proficiency in R.
His current focus lies at the intersection of clinical programming, statistical workflows, and Artificial Intelligence. He is passionate about understanding how AI can enhance productivity while ensuring human expertise, judgment, and critical thinking remain central to regulatory-grade clinical development.
Through this session, Avinash will demonstrate practical examples of Human-in-the-Control programming and provide attendees with actionable strategies for safely and effectively incorporating AI into their day-to-day programming activities.
