About Me

I am a Ph.D. candidate majoring in Social & Engineering Systems and Statistics at MIT IDSS and LIDS. I have been fortunately working with Prof. Marzyeh Ghassemi.

My current research centers around two primary topics:

• Understanding LLM Alignment: Characterizing when and why language models fail to use information effectively
  • Safety under multi-step, multilingual interactions (Speak Easy, ICML’25) and superficial style alignment (SafeStyle, ICLR’26)
  • Knowledge use and updates under retrieval augmentation (KScope, NeurIPS’25)
  • Intent Drift in long-horizon multi-agent systems (Ongoing Work)
• Improving LLM Alignment: Developing model- and system-level frameworks for language model adaptation
  • Supervised Fine-Tuning for instruction following (SFTMix, ACL’26)
  • Reinforcement Learning in healthcare (ETHOS-RL, In Submission)
  • Agent Harnesses for continual learning (ReHarness, In Submission)

For more information about my background and qualifications, please refer to my CV.

I am on the 2026–2027 industry job market and actively seeking full-time research roles. Please feel free to reach out if you think my background could be a good fit!