Dhruv Nathawani
Bio
Dhruv Nathawani is a Research Scientist at NVIDIA, where he works at the intersection of synthetic data and foundation model alignment. His work focuses on building data-centric methods that improve the reliability, adaptability, and performance of Nemotron foundation models and modern AI systems. Prior to NVIDIA, Dhruv was at Gretel, a synthetic data platform for developers, where he worked on tools for generating high-quality data for AI applications. At Salesforce Research, Apple Maps, and Carnegie Mellon University, Dhruv built AI systems spanning medical multimodal learning, document AI/OCR, satellite computer vision, and fMRI-based cognitive decoding.
Session (1)
Dhruv Nathawani is a Research Scientist at NVIDIA (previously Gretel, Salesforce, Apple) whose recent work spans mechanistic interpretability of LLMs, transferable activation-space interventions for AI safety, synthetic text-to-SQL datasets, and interpretability tooling for text-to-image models — making his session relevant for practitioners working on model steering, interpretability, and synthetic data.
Recent writing (4)
- Activation Space Interventions Can Be Transferred Between Large Language Models · paper · Mar 2025
- TinySQL: A Progressive Text-to-SQL Dataset for Mechanistic Interpretability Research · paper · Mar 2025
- DreamReader: An Interpretability Toolkit for Text-to-Image Models · paper · Mar 2026
- Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning · paper · Apr 2026
Public activity researched automatically · as of Jun 2026