Rahul Gupta

This account is verified. Learn more

@rahul1987iit

Responsible AI @Amazon | Organizer @TrustNLP |
Built Alexa, got Amazon Nova to be safe, now building next generation of Amazon models

guptarah.github.io/index.html Joined February 2011

308 Following 285 Followers 289 Posts

Latest posts 5

Pinned post

Rahul Gupta @rahul1987iit Oct 27

Our collaboration with OSU wins outstanding paper award at #EMNLP2026!!

The paper introduces, a benchmark for evaluating how vulnerable AI coding agents are to malicious "skill-based attacks". Where poisoned, reusable skill packages can compromise agents either immediately during a single task session or silently across sessions via self-mutating payloads.

Please see our paper here -- arxiv.org/html/2606.0254…

Huge shoutout to @yuting_ning, @Zhehao_Zhang123 and @hhsun1.

Rahul Gupta @rahul1987iit 4d

@trustnlp back for the 7th edition --

trustnlp @trustnlp 4d

After six editions across North America, from Seattle and Toronto to Mexico City and San Diego, the 7th edition travels to Asia, co-located with COLING in Macau, China.

📅 Papers due: January 5, 2027
Details: trustnlpworkshop.github.io

We would love to see your submissions.
View quoted post on X

Rahul Gupta @rahul1987iit Nov 2

The Trusted AI Symposium is back — this time in India 🇮🇳

📍 New Delhi | Feb 18, 2027
🤝 Hosted by Amazon AGI, AWS Public Policy & IIT Delhi
Details: tinyurl.com/25g9oc6j

For academia, industry & policy: building powerful AI that's inherently safe & reliable. This follows our engagement at the India AI Impact Summit and an ARA proposal focused on the Global South.

📄 Now accepting submissions: 4-page, non-archival papers on trusted AI.
⏳ Deadline: Nov 30, 2026

www.amazon.science Trusted AI Symposium 2027 Join leading researchers from India's academic institutions, industry, and policy community for an intensive, one-day symposium on the frontier of Trusted AI, held in New Delhi. Jointly hosted by...

Rahul Gupta @rahul1987iit Oct 27

Deadline around the corner. Looking forward to proposals -- specially on the Multilingual and Contextual AI Evaluation track.

Amazon Science Verified account @AmazonScience Oct 1

📣 Amazon Research Awards fall 2026 call for proposals is now open for submissions. Successful applicants will receive unrestricted funds, AWS promotional credits, and training resources.

Research areas:
• Agentic AI
• AI for Information Security
• Amazon Security
• Automated Reasoning
• AWS Cryptography
• Build on Trainium
• Multilingual and Contextual AI Evaluation
• On-Device Multimodal Perception
• Synthetic Data Generation for Multimodal Perception

Deadline: November 4
#AmazonResearchAwards

amazon.science/research-award…
View quoted post on X

Rahul Gupta @rahul1987iit Oct 24

Even rigorous internal safety testing benefits from external scrutiny. We partnered with @AVERIorg to independently review our safety evaluation datasets at Amazon; examining not just model outputs, but the quality and rigor of the tests themselves.

Blog at tinyurl.com/2u75xm49

Thanking @Miles_Brundage for a great collaboration.

www.amazon.science Strengthening AI safety evaluation through independent third-party review How Amazon partnered with AVERI to independently assess — and improve — our internal safety testing of Amazon models.