About
I work across AI systems, research engineering, and AI alignment. My current day-to-day work is on LLM post-training and alignment at Zoho Labs, and this site tracks the public edge of what I am building, studying, and trying to understand.
Current work
My technical work has ranged from database autotuning with reinforcement learning to LLM post-training, alignment, and evaluation. I am especially interested in how model behavior changes under training, what makes an evaluation actually faithful to the behavior we care about, and how tools reshape reliability when systems are deployed in settings where errors are expensive.
I also contribute to training and evaluation work for high-stakes domains like medical reasoning at MEDARC, where the core question is not just whether a model can answer, but whether it can be trusted.
Research collaborations
On the research side, I have worked with Robert McCarthy at UCL and with Lionel Levine and Jonathn Chang at Cornell. That work has focused on self-preservation propensity, emergent misalignment, and the side effects of character training, including constitutions and value drift in post-training.
Across these collaborations, the recurring theme is behavioral measurement: designing settings where the signal is real, the failure modes are legible, and the conclusions hold up outside the benchmark.
Programmable biology
I am also studying synthetic biology and bioinformatics, with a particular interest in the infrastructure that connects digital biological design to physical validation. More broadly, I want to understand how reliable AI systems can support programmable biology without collapsing under weak interfaces, bad measurements, or unclear feedback loops.
Open to hearing from
I am glad to hear from people working on AI alignment, model evaluation, tool use, generalization, medical AI, programmable biology, and technical fellowships or research programs.