Work
Selected work across adaptive AI systems, federated learning, distributed ML infrastructure, robust optimization, personalization, and reproducible experimentation.
Production Research Infrastructure
Adaptive federated personalization infrastructure for heterogeneous machine learning systems, resource-aware client routing, reproducible non-IID experimentation, policy evaluation, and serving-oriented research workflows.
Adaptive Inference Systems
Distributed adaptive inference infrastructure for reinforcement-learning-driven LLM routing, online optimization, observability, and production-scale serving experimentation.
Research Project
Researching geometry dynamics, update instability, and low-rank adaptation behaviour in federated learning under heterogeneous client data distributions.
Research Project
Studying client-specific personalization depth as a decision problem in federated learning, including head-only, partial, and full adaptation strategies under non-IID regimes.
Research Project
Exploring drift-aware adaptive aggregation and robustness mechanisms for federated learning under heterogeneous client partitions.