Kartik Mehta
| Kartik Mehta | |
|---|---|
| File:Kartik Mehta portrait.jpgKartik_Mehta_portrait.jpg | |
| Born | India |
| 🏳️ Nationality | Indian |
| 🎓 Alma mater | Indian Institute of Technology Delhi |
| 💼 Occupation | |
| Known for | Research on large language models, FLAMES, DeCRIM, DiCoRe |
| 🌐 Website | https://kartikmehta.me |
Kartik Mehta is an Indian computer scientist and Senior Applied Scientist at Amazon AGI, best known for his research on large language models, natural language processing, and generative artificial intelligence. He leads post-training and evaluation efforts for the Amazon Nova family of frontier models.[1][2]
Education
Mehta received his Bachelor of Technology and Master of Technology degrees in Machine Learning from the Indian Institute of Technology Delhi.[3]
Career and research
At Amazon AGI, Mehta works on advancing reasoning and alignment capabilities in large language models. His papers have appeared in venues such as the EMNLP and the NAACL.
Notable works include:
- FLAMES — Improving LLM Math Reasoning via Fine-Grained Analysis of the Data Synthesis Pipeline (EMNLP 2025).[4]
- DiCoRe — Enhancing Zero-Shot Event Detection via Divergent-Convergent LLM Reasoning (EMNLP 2025).[5]
- DeCRIM — LLM Self-Correction through Decompose, Critique, and Refine (EMNLP 2024).[6]
He holds a United States patent for an entity-extraction system deployed at scale within Amazon.[7]
Professional service
Mehta serves as a reviewer for major AI and NLP conferences including ACL, EMNLP, NAACL, ICML, and COLM. He has also served as a judge for Amazon’s research competitions such as the Alexa Prize and the Amazon Nova AI Challenge.[8]
Recognition
Mehta has been recognized as an IEEE Senior Member for professional excellence and contributions to applied AI research.[9]
External links
Category:Indian computer scientists
Category:Artificial intelligence researchers
Category:Amazon (company) people
Category:Living people
Category:IIT Delhi alumni
This article "Kartik Mehta" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:Kartik Mehta. Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
- ↑ "Amazon Nova models compete with ChatGPT, Google, Microsoft". Fortune. 3 December 2024. Retrieved 11 November 2025.
- ↑ "Amazon launches Nova Premier, its largest AI model yet". TechCrunch. 30 April 2025. Retrieved 11 November 2025.
- ↑ "IIT Delhi Alumni Directory". Retrieved 11 November 2025.
- ↑ Seegmiller, Parker; Mehta, Kartik; Saha, Soumya; Tao, Chenyang; Oraby, Shereen; Gupta, Arpit; Chung, Tagyoung; Bansal, Mohit; Peng, Nanyun (2025). FLAMES: Improving LLM Math Reasoning via a Fine-Grained Analysis of the Data Synthesis Pipeline. ACL Anthology. pp. 24746–24766. doi:10.18653/v1/2025.findings-emnlp.1346. ISBN 979-8-89176-335-7. Retrieved 11 November 2025. Search this book on
- ↑ Parekh, Tanmay; Mehta, Kartik; Mehrabi, Ninareh; Chang, Kai-Wei; Peng, Nanyun (2025). DiCoRe: Enhancing Zero-Shot Event Detection via Divergent-Convergent LLM Reasoning. ACL Anthology. pp. 20571–20593. doi:10.18653/v1/2025.emnlp-main.1038. ISBN 979-8-89176-332-6. Retrieved 11 November 2025. Search this book on
- ↑ "LLM Self-Correction with DeCRIM". ACL Anthology: 7773–7812. November 2024. doi:10.18653/v1/2024.findings-emnlp.458. Retrieved 11 November 2025.
- ↑ "US Patent 12,190,063: Machine Reading Comprehension-Based Named Entity Recognition". Retrieved 11 November 2025.[permanent dead link]
- ↑ "Amazon Nova AI Challenge". Retrieved 11 November 2025.
- ↑ "IEEE Senior Member Program". Institute of Electrical and Electronics Engineers (IEEE). Retrieved 11 November 2025.
