ANURAG TRIVEDI

Designed and deployed production-ready agentic RAG workflows and dual-LLM regulatory validation, improving query accuracy by 30% and reducing hallucinations by 25% while scaling semantic search to 10,000+ docs with <1.1s p95 latency.

Built biomedical image analysis and knowledge-graph pipelines that improved condensate segmentation accuracy by 10% and enabled 3× faster therapeutic pathway exploration via GraphRAG integrations.

Built a GraphRAG system for querying project specifications, reducing response times by 40% and designing schemas to model regulatory relationships for compliance-aware retrieval.

Developed deep-learning data-compression systems achieving 50% size reduction with 99% accuracy and optimized GPU-cluster deployments to improve resource utilization by 35%.

Designed a Multi-Gate Mixture-of-Experts model achieving 25% accuracy gain with 30% less training time and developed explainable AI methods to improve traceability and reduce overfitting risk by 15%.

Shipped ML-driven product features and scalable data pipelines for >50M records, increasing user engagement by 15% and improving pipeline accuracy by 30% while reducing processing time by 40%.

M.Sc. in Applied Mathematics with a thesis on a cubic B-spline collocation method for parabolic partial differential equations.

B.Sc. (Hons.) in Mathematics with foundation coursework in linear algebra, probability, statistics, and optimization.

M.Sc. in Computational Modeling and Simulation with a thesis on multi-task learning for holistic data-driven models of engineering systems.