Eshan Jain

AI Researcher

Gurugram, Haryana, India1 yr 7 mos experience

Key Highlights

Developed GNNXemplar framework for GNN interpretability.
Presented research at NeurIPS with high acceptance rate.
Strong foundation in AI and quantitative finance.

Stackforce AI infers this person is a Quantitative Researcher with a strong focus on AI and finance.

Contact

Skills

Core Skills

Machine LearningQuantitative Research

Other Skills

Alpha GenerationC++Cascading Style Sheets (CSS)ChessCommunicationComputer ScienceComputer VisionData ScienceDeep LearningEngineeringFinanceFinancial MarketsGraduate LevelGraphic DesignHTML5

About

I'm Eshan, a Quantitative Researcher at Graviton, working at the intersection of high-frequency trading and AI. A recent Computer Science & Engineering grad from IIT Delhi, I blend a strong foundation in programming and math with a deep interest in finance and machine learning. My journey through IIT Delhi and now at Graviton has only amplified my passion for unraveling complex algorithms, harnessing technology to solve challenging problems, and the endless possibilities they unlock. I thrive on curiosity, whether it's exploring the fascinating world of quantitative finance or diving deep into the state-of-the-art breakthroughs in AI, I’m always eager to learn and grow. The thrill of solving difficult problems, and staying on the cutting edge of tech is what drives me every day. If you share a love for tech, finance, or just enjoy discussing the latest advancements, let’s connect!

Experience

Graviton research capital llp

Quantitative Researcher

Jun 2025 – Present · 9 mos · Gurugram, Haryana, India · On-site

Indian institute of technology, delhi

AI Researcher

Aug 2024 – Jun 2025 · 10 mos · Delhi, India

My work here culminated in a paper “GNNXemplar: Exemplars to Explanations—Natural Language Rules for Global GNN Interpretability” being accepted for an oral presentation at NeurIPS 2025 (77 selected out of 21,575 submissions).
Brief about our work: In this work, we introduce GNNXemplar, a framework that explains GNNs through exemplar nodes in the embedding space and translates their neighborhood patterns into natural language rules using large language models. This shift from motif discovery to exemplar-based reasoning leads to explanations that are more faithful, scalable, and human-interpretable—as also validated by a user study with 60 participants.

Large Language Models (LLM)Machine LearningResearchQuantitative ResearchData Science