Jan 2025
Building Abstention-Aware RAG Systems
Strategies to make RAG systems recognize uncertainty in high-stakes domains.
Ranjan Mondal
Senior Associate, Applied AI/ML at JP Morgan Chase & Co.
PhD in Machine Learning (ISI Kolkata)
I am a Senior Associate in Applied AI/ML at JP Morgan Chase, where I lead initiatives in large-scale graph learning, LLM orchestration, and financial optimization. With a PhD in Machine Learning from the Indian Statistical Institute and over five years of industry experience, I specialize in bridging cutting-edge research with scalable business solutions. My expertise spans LLM fine-tuning (SFT, RLHF, DPO), Retrieval-Augmented Generation (RAG), Graph Neural Networks, and Reinforcement Learning.
News
Feb 2025
Talk
Delivered invited talk on "LLM Fine-Tuning: From Transformers to DPO" at National Workshop on LLMs & Agentic AI, Chandigarh University.
Jan 2025
Award
Received Impact and Innovation Award at JP Morgan Chase for Retrieval-as-a-Service platform.
Jul 2024
Talk
Invited speaker at SERB Workshop on "Deep Learning: Foundation to Cutting Edge Technologies", UPES Dehradun.
Experience
Senior Associate, Applied AI/ML
JP Morgan Chase & Co.
2023 – Present
Leading development of Retrieval-as-a-Service platform, Agentic RAG systems for legal contracts, Graph RAG implementations, and LLM fine-tuning initiatives.
Lead Research Engineer
Samsung Research
2021 – 2023
Worked on video panoptic segmentation for embedded platforms with real-time instance tracking and temporal stability.
Research Areas
LLMs
Fine-tuning, RLHF, DPO, vLLM serving
Graph ML
GNNs, Knowledge Graphs, Graph RAG
RAG Systems
Hybrid retrieval, Agentic RAG
Quant Trading
Algorithmic strategies, Backtesting
Publications
Blog
Nov 2024
From SFT to DPO: LLM Fine-Tuning Guide
Comparing alignment techniques for domain-specific models.