Dipayan Dey

About Dipayan Dey

I am a passionate data professional with a strong academic foundation in Statistics…

Experience

1. Associate

   Goldman Sachs

   Nov 2024 — Present · Bengaluru, IN

   At Goldman Sachs, I worked at the intersection of Large Language Models (LLMs), regulatory compliance, and quantitative finance, designing AI architectures that are interpretable, regulation-safe, and enterprise-scalable- LLM Architecture & Semantic SearchBuilt Matryoshka Loss–based LLM systems for semantic and hybrid search, enabling contextual retrieval across regulatory filings, audit data, and compliance logs. Optimized for low-dimensional embeddings and explainable reasoning in sensitive financial environments- Multi-Agent RAG FrameworksDeveloped LangGraph–LangChain–LlamaIndex architectures for multi-agent Retrieval-Augmented Generation (RAG). Designed supervisor–agent orchestration enabling real-time analysis of Earnings Calls, Market Data, and Regulatory Reports with compliance-grade accuracy- Compliance & Market SurveillanceValidated and enhanced Pump-and-Dump and Intra-Market Manipulation models using regression, ML classifiers, and hybrid benchmarking. Created explainable frameworks for market surveillance and regulatory audit alignment- Regulatory Forecasting & Credit ModelingWorked on CCAR & ICAAP PPNR models for Mortgage-Backed Products, projecting 9- and 12-quarter revenues under stress scenarios.Built and calibrated Credit Card Payment Rate and Revolve Rate models, improving behavioral forecasting and credit risk management- Conversational AI & Auto-SummarizationEngineered Auto-Summary and Auto-Compose systems using fine-tuned LLMs to summarize analyst–client interactions and assist human analysts with compliant, context-aware replies- Regulation 23B FrameworkDeveloped compliance architecture to govern transactions between the parent bank and subsidiaries, ensuring traceability, transparency, and rule adherence- Market Risk – Global Market Shock (GMS)Calibrated firmwide risk factor shocks for systemic market stress scenarios, linking quantitative risk models with AI-driven interpretability.

Education

• Vivekananda Mission High School

  School, Science

  2009 — 2016

• Indian Institute of Technology, Kanpur

  MSc, Statistics

  2019 — 2021

• Ramakrishna Mission Residential College

  Bachelor's degree, Statistics

  2016 — 2019