Agentic AI, biology foundation models, and GPU-native chemistry are converging to rewrite drug discovery. The Giri Pillai Lab works at that frontier - from autonomous co-scientists to longevity therapeutics.
Autonomous LLM agents (GPT-class, Claude, open Llama/Mistral) that plan experiments, call docking/MD tools, critique their own hypotheses, and iterate on molecular design with minimal human steering. The lab partner of 2026–2036.
AlphaFold 3, Boltz, RoseTTAFold All-Atom, ESM, and Chai-1 reshaping structural biology. Fine-tuned protein & molecule foundation models are becoming the new substrate for SAR, ADMET, and target discovery.
Diffusion models, flow-matching, and equivariant generative networks design novel scaffolds with built-in synthetic accessibility - pushing de novo design from curiosity to production tool over the next decade.
Hallmarks-of-ageing targets, senolytics, partial reprogramming, and AI-driven biomarker discovery. By 2036, computational longevity moves from research curiosity to credentialed therapeutic modality.
GPU-accelerated MD (OpenMM, GROMACS-CUDA), neural network potentials (MACE, NequIP), and early quantum-assisted free-energy methods - compressing physics-grade simulations from weeks to hours.
KNIME, Jupyter, and modern MLOps stacks (W&B, MLflow, Hugging Face) keeping discovery transparent and reproducible - open datasets for SARS-CoV-2, tyrosinase inhibitors, and longevity targets.
A pragmatic mix of open-source cheminformatics, proprietary modelling suites, and modern ML - chosen per problem.
Filter the lab's active work — from agentic co-scientists to diffusion chemistry, longevity, and GPU-native simulation. Click a topic or search the stack.
LangGraph-orchestrated LLM agent that proposes analogues, runs docking + ADMET tools, critiques its own hypotheses, and returns ranked design rounds with audit trails.
Foundation-model pipeline for complex prediction across protein–ligand–nucleic systems, feeding structure-aware ML scoring into early-stage lead identification.
Equivariant diffusion and flow-matching generators with built-in synthetic-accessibility filters — pushing de novo chemistry from curiosity to production tool.
ESM-3 + cell-state foundation models mapped against hallmarks-of-ageing targets, with senolytic and partial-reprogramming triage informed by AI biomarker discovery.
OpenMM + MACE / NequIP neural-network potentials compressing physics-grade free-energy calculations from weeks to hours on commodity GPUs.
Early integration of quantum-assisted free-energy methods with classical MD to push accuracy boundaries on tough congeneric series.
Curated, reproducible SAR & docking datasets released with KNIME / Jupyter notebooks so the community can rebuild every model end-to-end.
ML-driven SAR modelling for vector-borne disease control — combining classical descriptors with modern molecular representation learning.
Agentic system reading longitudinal ageing biomarkers and literature to propose mechanism-anchored intervention arms for healthspan studies.
Showing 9 of 9 projects · Horizons reflect the 2026–2036 lab roadmap.