Lawrence T. Lin
PhD Candidate in Physics | Cornell University
I am a PhD candidate in Physics at Cornell University, specializing in experimental cosmology and low-temperature instrumentation under Professor Mike Niemack. My research focuses on developing advanced machine learning techniques for time-ordered data analysis from cosmological telescopes, with applications in the Simons Observatory and the CCAT-Prime telescope collaboration. I work on optimizing ultra-efficient neural networks mapped directly to hardware and exploring novel initialization techniques to overcome deep learning challenges. I have extensive experience with detector characterization, cryogenic systems, thermal modeling, and data analysis in the context of mm/sub-mm astronomy. My skills span physics, data science, machine learning, hardware optimization, and thermal engineering.
Experience
Graduate Research Assistant
Cornell University
February 2022 - Present
- Lead thermal design, testing, and validation for CCAT-Prime/FYST Mod-Cam cryogenic instrument, achieving temperature stability of 0.1K/day at 40K stage and 0.2mK at 100mK stage
- Developed machine learning models for detector data quality assessment and transient signal identification for the Simons Observatory (SO) and CCAT-Prime, improving data processing efficiency
- Conducted extensive cryogenic testing of detector arrays, including optimization of readout parameters, characterization of detector noise properties, and analysis of optical loading
- Designed and executed experiments to measure thermal conductivity of materials at cryogenic temperatures, informing cryostat design decisions
- Established detector matching algorithms between resonant frequencies and physical detector positions for kinetic inductance detectors (KIDs)
- Implemented and managed cryogenic monitoring systems using Python, Grafana, Docker, and OCS, enabling remote operation of complex cryogenic systems
Projects
Machine Learning for Detector Quality Assessment
2023-Present
Developing transformer-based models to identify "bad" detectors from time-ordered data, enabling automatic data quality assessment. This model incorporates spatial and frequency-domain information to detect various detector failure modes and readout issues for both transition edge sensor (TES) and kinetic inductance detector (KID) technologies. Implemented using PyTorch, the system is designed to scale to handle tens of thousands of detectors with minimal human intervention.
Mod-Cam Cryogenic System Design and Testing
2022-Present
Led the thermal design and testing of Mod-Cam for CCAT-Prime/FYST, a pathfinder instrument with sub-100mK operating temperature. Conducted extensive thermal modeling, including FEA simulations and analytical calculations to optimize heat strap designs. Characterized thermal conductivity of different copper alloys through RRR measurements. Achieved temperature stability of 0.1K/day at 40K and sub-mK stability at 100mK, enabling high-sensitivity detector characterization.
Detector Matching Algorithm for Kinetic Inductance Detectors
2024-Present
Developed algorithms to match resonant frequencies with physical detector positions for KID arrays, accounting for frequency shifts due to optical loading and bath temperature changes. This work is critical for astronomical observations as it enables precise mapping between detector signals and sky positions, improving map-making procedures and astronomical data quality.
Remote Cryogenic Operation Systems
2023-Present
Designed and implemented monitoring and control systems for remote operation of complex cryogenic systems, including dilution refrigerators at high-altitude astronomical sites. Developed Python-based monitoring tools with Grafana visualization and alert systems. Created reliable protocols for remote operation, power recovery, and emergency handling of cryogenic equipment.
Optical Characterization of Ultra-High Frequency Detector Arrays
2022-Present
Conducted comprehensive optical testing of detector arrays for the Simons Observatory, characterizing noise properties, optical efficiency, and saturation power. Analyzed data to validate detector models and inform future design improvements. Developed specialized cold load testing procedures for precise calibration of mm/sub-mm detectors.
Publications & Presentations
For a complete list of my publications and collaborative work, please see my Google Scholar profile.
Skills
Contact
Feel free to reach out if you'd like to connect about research opportunities, collaborations, or discussions about physics, data science, machine learning, and hardware innovation.
Email: ltl32@cornell.edu Copy to clipboard
LinkedIn: linkedin.com/in/lawrencetlin
Google Scholar: scholar.google.com