About Me
I'm a Member of Technical Staff at Liquid AI. I work on research and engineering problems across the foundation model stack: architecture design, pretraining, distillation, and model compression.
I earned my master's and bachelor's degrees in Computer Science from Stanford University. My technical interests span foundation model development and building AI-native applications.
In my free time, I enjoy golfing, skiing, and documenting life — whether through a camera or through text.
Reach me at [firstname][lastname]@gmail.com.
Work
Quantization-Aware Distillation for 4-Bit On-Device Language Models
Quantization-aware distillation for 4-bit LFM2.5 models, recovering most of the quality lost to post-training quantization at Q4_0 memory and speed.
NeurIPS 2026 Workshop on On-Device Intelligence (Oral)
Blog with Leonie Monigatti
In-Place Tokenizer Expansion for Pre-trained LLMs
An in-place recipe for expanding a pre-trained model's tokenizer while preserving source-checkpoint quality.
arXiv preprint, 2026
LFM2 Technical Report
Technical report for Liquid Foundation Models 2.
arXiv preprint, 2025
33 authors in alphabetical order
Simple, Scalable Reasoning via Iterated Summarization
A method for scaling language model reasoning over long contexts.
ICML 2025 Workshop on Long Context Foundation Models
ICML 2025 Workshop on AI for Math
* Equal contribution