I am a Postdoctoral Research Associate in the Vlachos Research Group · Predictive Science Lab at Purdue University, working on the Lilly–Purdue Research Collaboration.
My research builds physiologically-grounded digital twins of human health and disease—predictive models that translate complex physiology into clinical insight and de-risk drug development. In parallel, I develop AI agents that accelerate how these models are constructed, calibrated, and refined.
What I work on
- Whole-body digital twins. Closed-loop, lumped-parameter cardiovascular and lymphatic transport simulators that capture organ-level hemodynamics, interstitial pressure, and route-of-administration physiology.
- PBPK / PK-PD modeling. Compartmental ODE systems for hepatic clearance, oral and subcutaneous absorption, and CNS drug delivery; in vitro–to–in vivo extrapolation (IVIVE) for human-relevant predictions.
- Calibration & inference. Variational inference, Bayesian and gradient-based parameter estimation, sensitivity analysis, and differentiable simulation in Python/JAX.
- AI agents for model authoring. A domain-specific language for compartmental PK models that lets agents reliably build, validate, and iterate on PBPK structures within the Design–Make–Test–Analyze loop.
Background
I completed my Ph.D.\ in Computer Science at IIT Kharagpur (2024) on “Application of computational modeling towards digital twinning of human physiological and biochemical processes,” supported by India’s competitive SERB Overseas Visiting Doctoral Fellowship at Purdue (2023–2025). Before that, I held research and software-engineering roles spanning industrial PBPK modeling (esqLABS GmbH), metabolic systems (Metflux Research), and embedded systems (Embedon Global Energy).
Get in touch
The fastest way to reach me is by email: maravind@purdue.edu. I’m also on LinkedIn and GitHub.