- Multimodal Enrollment Forecasting : Build hierarchical models that forecast "Top-Down" (Country/Study) and "Bottom-Up" (Site-level) enrollment by ingesting real-time screening logs from IRT and site-activation milestones from CTMS.

- Discontinuation & Attrition Modeling : Implement Survival Analysis (Cox-PH, DeepSurv) and RNNs to predict patient dropout probability using longitudinal data from EDC, which serves as the primary driver of "Maintenance Phase" demand.

- Demand vs. Supply Optimization : Develop Monte Carlo simulations or Stochastic Optimization models to determine safety stock levels, balancing the variance between predicted enrollment and actual inventory on hand.

- Dose Titration Logic : Build predictive ML models to anticipate dose escalations or reductionssyncing with IRT dispensing data to ensure the correct kit strength is available at the site before the patients next visit.

- Clinical Data Lake Management :

a. Architect unified data pipelines that join EDC (clinical outcomes/visit data) with IRT (supply/randomisation data).

b. Manage the full ML lifecycle (Tracking, Registry, Serving) to ensure model reproducibility.

c. Build resilient, real-time pipelines for monitoring supply-demand signals and triggering automated alerts for potential stock-outs.

Required Technical Expertise :

- Systems Integration : Proven experience processing and feature-engineering data from EDC (e.g., Medidata Rave, Veeva) and IRT/RTSM platforms.

Advanced ML Domains :

- Time-Series : DeepAR, Temporal Fusion Transformers (TFT), or N-BEATS for non-linear recruitment trends.

- Survival Analysis : Expert-level experience modeling "Time-to-Event" data to handle censored patient discontinuation patterns.

- Probabilistic Programming : Experience with PyMC or Gurobi/OR-Tools to solve the "Supply vs. Demand" constraint problem.

- Data Engineering : Expert-level Python, SQL, and distributed computing for processing large-scale, high-velocity clinical datasets.

Clinical Domain Knowledge (Preferred to have) :

- Clinical Systems : Deep understanding of the data schemas within IRT/RTSM (Randomisation/Dispensing) and EDC (Patient Visits/Adverse Events).

- Supply Dynamics : Understanding of "Initial Seeding," "Trigger-based Resupply," and "Dose Titration" within a global trial context.

- Regulatory Context : Experience working within GxP / CFR Part 11 compliant environments, ensuring model auditability.

- Standards : Knowledge of CDISC (SDTM/ADaM) data structures is a significant plus