partha majumder

Architected and shipped 10+ production-grade GenAI platforms featuring Perplexity-style research engines, real-time SSE streaming, and event-driven media pipelines to improve reliability and scaling of LLM-driven products.

Delivered high-performance numerical simulation systems and accelerated stochastic particle simulations (3M+ particles) using CUDA, achieving 110x speedup versus CPU and production-grade reliability.

Built RAG and agentic AI prototypes with LangChain and FastAPI, optimized LLM inference and vector retrieval to reduce latency by 40%, and shipped full-stack AI integrations for secure data persistence.

Delivered NLP-based recommendation and search systems for startup intelligence, including investor matching and similar-company discovery using BERT and SpaCy.

Developed hybrid deep learning and metaheuristic optimization frameworks (ELM + evolutionary algorithms) and engineered scalable NLP pipelines for large-scale Twitter analysis.

Applied deep learning for groundwater contaminant source identification and developed hybrid simulation–optimization workflows for aquifer parameter estimation and remediation strategy design.

Enhanced groundwater management software with scalable optimization workflows and integrated deep neural networks and surrogate-based optimization to accelerate remediation planning.

Enhanced enterprise net-banking applications using Java/J2EE and Oracle, improving framework features, performance, and maintainability.

Ph.D. in Computational Geoscience focused on simulation–optimization frameworks combining neural networks and metaheuristic algorithms to accelerate high-dimensional inverse modeling.

Grade: 9.26/10

M.Tech in Water Resources Engineering with work on simulation–optimization frameworks and numerical solvers for constrained multi-objective optimization.

B.Tech in Civil Engineering covering core civil engineering principles and computational methods applied to water resources and environmental systems.