Shovon Sengupta ^{Top-Executive}

Over the past 18 years, I have worked at the intersection of applied machine learning research and enterprise AI — exploring how rigorous statistical and deep learning methods can be translated into systems that are not only performant, but interpretable, trustworthy, and consequential.As Director of the AI Center of Excellence at Fidelity Investments, I lead teams working on some of the more interesting and difficult problems in applied AI — from grounding large language models in complex financial document workflows, to designing reinforcement learning frameworks for personalization, to building explainability architectures that make deep learning models auditable in regulated environments.Research & Applied Interests:- Generative AI & LLMs — Fine-tuning, DPO optimization, agentic workflows, retrieval-augmented generation, and the reasoning boundaries of foundation models in domain-specific contexts- Explainable & Trustworthy AI — Architect of FidxAI, a proprietary XAI framework operational across wide range of production models; my broader research interest lies in making complex models legible to both technical and non-technical stakeholders- Graph Machine Learning & Reinforcement Learning — Knowledge graph-powered recommender systems, contextual bandits, and self-learning routing frameworks; work in this space presented at KDD '23- Time Series & Macroeconomic Forecasting — Application of Neural ARFIMA, LLM-based, and wavelet methods to financial and macroeconomic forecasting under uncertainty; published in the International Journal of Forecasting (Elsevier) and Computational Economics (Springer Nature)- Causal Inference & Uplift Modeling — Causal pattern mining and treatment effect estimation for high-stakes business decisions- NLP & Prompt Engineering — Research on prompting strategies for non-reasoning LLMs; ongoing work on LLM-based drift detection and document question answeringConferences & Speaking:- International Symposium on Forecasting (Oxford '22 · UVA '23) · AI4 Las Vegas '23 · Federal Forecasters Conference, U.S. Department of Treasury '24Teaching & Community:- Panelist, Bytes of Wisdom — Questrom School of Business, Boston University- Founder, Fidelity Data Science BootcampEducation:- Ph.D., BITS Pilani (Joint research collaboration with Sorbonne University, Paris) — Thesis: Application of Machine Learning in Finance- MBA, Questrom School of Business, Boston University (Expected 2028)Finally, I am always open to thoughtful conversations around research, enterprise AI challenges, speaking opportunities, and academic collaboration.