Aditya Challamarad

Experience

1. Machine Learning Researcher

   University At Buffalo School Of Engineering And Applied Sciences

   Jun 2024 — Present · Buffalo, NY, US

   Co-authored GloVoAD, a multilingual ADRD dataset (n=2,068; English, Spanish, and Mandarin), enriched with cognitive scores and demographic metadata.• Developed a language-agnostic detection pipeline using WhisperX/Pyannote for diarization and OpenSMILE for acoustic features extraction, without relying on transcripts.• Trained a binary SVM on ComParE 2016 (6,373-dim) features with hyperparameter tuning and Relief-F feature selection, achieving 0.7247 F1 in-domain and 0.80 F1 on external multilingual corpora by adjusting threshold.• Designed a two-stage classifier for ADRD subtype prediction (AD, MCI, PPA, Probable AD) using ExtraTrees on Emobase feature, improving test F1 from 0.781 to 0.803 using feature importance.• Built an EEG Classification pipeline using MNE and PyTorch lightning with EEGNet to classify experimental (stimulus) and neutral participants. Applied windowing, group based splitting, and weighted training for improved accuracy.• Led NIH replication prize (track one) proposal on AI voice biomarkers for neurological and cognitive disorders, explaining the scientific rationale, public impact and a plan for replication across multiple datasets, languages, and recording conditions.

Education

• GEMS Education

  High School Diploma, A levels

  2018 — 2022

• University at Buffalo

  BSc Computer Engineering, Minor: Electrical Engineering