News & Accolades

Co-organized TrustNLP @ ACL'26. ACL papers: LLM Unlearning, SWAN, ARES, and AutoSUIT Bench.

Presented Amazon's work on Frontier Safety in AI at IIT Bombay and IIIT Delhi as part of the India AI Summit. Amazon signed the "Frontier AI Commitments" at the Summit; my team is responsible for building the technical foundation for Amazon's AI offerings under these commitments.

Amazon Science article featuring my team's work: Building trust into AI.

Amazon Science blog: How catastrophic is your LLM?

Red Hat blog: AI trust through open collaboration: A new chapter for responsible innovation.

Organized the Trusted AI Symposium 2026 in New York City, an invitation-only event bringing together academic and industry researchers on AI safety and governance.

Collaborated on three papers: LLM Deception, Catastrophic Risk, and Jailbreak Generalization.

Winner of the AgentX Benchmark & Evaluation Track for Establishing Best Practices for Building Rigorous Agentic Benchmarks.

Co-organized the SemEval Unlearning Challenge at ACL.

My team leads the Nova AI Challenge, supporting 10 universities annually on safety research.

Represented Amazon at the Frontier Model Forum and Partnership on AI.

Papers: VMDT, Agentic Benchmarks, and LLM Unlearning (workshop).

Paper: Localizing and Deleting Toxic Memories in LLMs (Findings).

Paper: Quantitative Certification of Bias in LLMs.

Papers: MUNCH and K-Edit (Preventing and Detecting LLM Misinformation workshop).