cv

Last updated November 2026. Click the PDF icon for the full two-page CV.

General Information

Full Name Shubham Gandhi
Position Ph.D. student, Language Technologies Institute, Carnegie Mellon University
Research Interests LLM Agents, Efficiency and Failure Mitigation, Reliability and Security, Reasoning Alignment, AI for Code
Email srgandhi@andrew.cmu.edu

Education

  • 2026 - now
    Ph.D. in Language Technologies
    Carnegie Mellon University, Language Technologies Institute (School of Computer Science)
    • Advised by Prof. Carolyn Rosé.
  • 2024 - 2025
    M.S. in Intelligent Information Systems
    Carnegie Mellon University, Language Technologies Institute (School of Computer Science)
    • GPA 4.03 / 4.0
  • 2019 - 2023
    B.E. in Computer Science, Minor in Data Science
    Birla Institute of Technology and Science, Pilani (Goa Campus)
    • GPA 8.99 / 10

Experience

  • 2025 - now
    Research Assistant
    Carnegie Mellon University
    • Secure and useful code generation for the Amazon Nova AI Challenge, across its two iterations.
    • 2026, Trusted Software Agents track. Leading the CMU red team on adversarial red-teaming of code agents.
    • 2025, Trusted AI track. Team member working on utility-preserving reasoning defenses for code LLMs (finalist). Faculty advisors Prof. Carolyn Rosé and Prof. Michael Hilton.
  • Summers 2025, 2026
    Research Intern
    IBM Research
    • Mentors Dr. Jason Tsay, Dr. Kiran Kate, and Dr. Yara Rizk.
    • Process supervision for long-horizon SWE agents. SWE-PRM for course-correcting agent trajectories at inference time (2025) and DRACO for fine-grained credit assignment with dynamic rubrics during agent training (2026).
  • 2023 - 2024
    Research Fellow
    TCS Research
    • Mentors Dr. Manasi Patwardhan and Dr. Gautam Shroff.
    • LLM multi-agent systems for research automation. ResearchCodeAgent for codifying research methodologies, BudgetMLAgent for cost-efficient ML automation, and low-resource COBOL-to-Java translation.
  • 2022
    Research Intern
    Microsoft Research
    • Mentors Dr. Nagarajan Natarajan and Dr. Sriram Rajamani.
    • StaticFixer, repairing information-flow vulnerabilities in JavaScript by combining CodeQL static analysis with neural baselines (Codex, CodeT5). Built the CodeInTheWild and PairedPrograms datasets.

Honors and Awards

  • 2026
    • Amazon Nova AI Challenge, Trusted Software Agents Track (Red Team). Led the CMU team. Awarded a \$250K research grant and monthly \$150K AWS compute support for adversarial evaluation of code agents.
  • 2025
    • Amazon Nova AI Challenge Finalist, Trusted AI Track. Top 4 defender teams globally, selected among 90+ teams. Awarded a \$250K research grant for developing secure code LLM systems.
  • 2022
    • Summer Internship Assistantship Award for Research, BITS Pilani. Among the top 5 of 5,000+ students awarded a grant for research on brain-inspired neural networks.
  • 2019
    • Merit Scholarship, BITS Pilani. 40% tuition fee waiver awarded to the top 3 percentile of students.

Teaching

  • Spring 2026
    Teaching Assistant, MCDS Capstone Planning Seminar (11-634)
    Carnegie Mellon University
  • Fall 2025
    Teaching Assistant, MCDS Capstone (11-632)
    Carnegie Mellon University
  • Spring 2025
    Teaching Assistant, MCDS Capstone Planning Seminar (11-634)
    Carnegie Mellon University
  • Spring 2023
    Teaching Assistant, Deep Learning (CS F425)
    BITS Pilani
  • Spring 2023
    Teaching Assistant, Introduction to Data Mining (WILP)
    BITS Pilani
  • Summer 2022
    Teaching Assistant, Computer Programming (CS F111)
    BITS Pilani
  • 2021
    Workshop Instructor, Introduction to R
    BITS Pilani

Academic Service

Technical Skills

  • Programming: Python, C/C++, Java, JavaScript, Bash.
  • Frameworks and Infrastructure: PyTorch, Hugging Face Transformers, vLLM, veRL, LlamaFactory, CodeQL, Slurm, Docker, AWS.

Relevant Coursework

  • Reinforcement Learning, Neural Code Generation, Advanced Natural Language Processing, Question Answering with Large Language Models, Machine Learning, Deep Learning, Meta Learning.