Aman Choudhri
Statistics & math graduate; probabilistic ML researcher and software engineer
aman
New York, New York, USA
Joined December 2024
Network
1.7K connectionsAMNIKD
GAGRMY
AKSBHY
TMGGCC
NCAAAC
YBSHSH
JRGBCC
Summary
Academic researcher with strong probabilistic and computational-ML skills applied to neuroscience and audio: coauthor on eLife paper and ML-for-audio/NeurIPS work using VAEs, mixture models, and uncertainty quantification to analyze hundreds of thousands of rodent vocalizations and to localize calls from multi‑channel audio. nih+1
Practical ML + infrastructure practitioner: maintains public repos for applied projects (blogregator, windopt, fema-assist), implements SLURM orchestration and multi-fidelity optimization, and lists ML ops / infra as current focus—indicating ability to move ML research tools into scalable production workflows. github+3
Bridges quantitative research and public policy: conducted a large YouGov-linked survey project in Professor Andrew Gelman’s lab exploring social networks and political preferences, and interned at Public Policy Lab—showing interest and experience in applying statistical methods to policy questions. columbia+1
Student leader and communicator: served as captain/president of Columbia Bhangra, managing a 30-person team and sizeable budget while also engaging in research and software projects, indicating leadership, event coordination, and cross-functional communication skills. columbia+1
Work
Education
Projects
Writing
Unsupervised discovery of family specific vocal usage in the Mongolian gerbil
January 1, 2024Paper applying unsupervised deep learning (VAE + clustering) to 583k gerbil vocalizations recorded continuously, showing stable, family-specific vocal 'dialects' and sequence structure; positions gerbils as a model for vocal-communication neural studies.
Vocal Call Locator Benchmark (VCL) for localizing rodent vocalizations from multi-channel audio
January 1, 2024Benchmark and modelling work for sound-source localization and vocalization attribution in multi-channel recordings; presented at ML-for-Audio venues (NeurIPS workshop/materials).
Effects of aggressive and conservative strategies for mechanical ventilation liberation
January 1, 2023Clinical/biostatistical work on strategies for ventilator liberation; coauthored with clinical researchers (Journal of Critical Care).