Scholar
Atharva Naik
Google Scholar ID: wTTF4yYAAAAJ
PhD Student, Carnegie Mellon University
Citations & Impact
Citations
1,335
H-index
7
i10-index
6
Publications
20
Co-authors
9
list available
Publications
Resume (English only)
Academic Achievements
- Project MetaLint: Training LLMs to flag non-idiomatic Python and Java code while adapting to evolving best practices; PBEBench: A dynamic, contamination-free, and scalable synthetic benchmark for evaluating knowledge-free inductive reasoning through multi-step string rewrites; CRScore: A reference-free metric for evaluating code review comments using LLMs and static analysis; Led the CMU team to the finals of the Amazon Nova AI Challenge in 2025, training LLMs to generate secure code; Explored using LLMs to generate reflection questions in collaborative SQL programming activities, with a study accepted at AIED 2024 and nominated for Best Paper and Best Student Paper.
Research Experience
- Research Assistant at Carnegie Mellon University (Jan 2023 - present); Research Assistant Intern at Technische Universität Darmstadt (Aug 2021 - Dec 2021); Research Assistant Intern at University of Alberta (Apr 2021 - Sep 2021); Research Intern at Adobe, Bangalore (May 2021 - Aug 2021); Undergraduate Student Researcher at Autonomous Ground Vehicle Research Group, IIT Kharagpur (Feb 2019 - Mar 2020).
Education
- PhD in Language Technologies, Carnegie Mellon University (incoming Aug 2024); Masters in Language Technologies, Carnegie Mellon University (Aug 2022 - May 2024, GPA: 4.14/4); Bachelor of Technology, Computer Science and Engineering, Indian Institute of Technology, Kharagpur (July 2018 - May 2022, GPA: 9.66/10).
Background
- A second-year PhD student at the Language Technologies Institute (LTI), Carnegie Mellon University, advised by Prof. Carolyn Rose. Research focuses on synthetic data generation and post-training/self-improvement methods for enhancing LLM-generated code quality, reasoning, and code safety/alignment.
Miscellany
- Personal interests and hobbies not mentioned in the provided HTML content.
