Selected Work & Research
A curated overview of applied machine learning systems, real-world experience, and technical deep dives.
Biomedical ML · Noisy data · Robust evaluation · Real-world systems
End-to-end ML for biomedical imaging under noisy, real-world conditions: preprocessing, validation, error analysis, and tradeoffs when data and deployment constraints are tight.
Selected deep dives
How systems were built, debugged, and evaluated in practice: concise technical case studies.
Project walkthroughs
Short walkthroughs of systems built and evaluated in practice.
End-to-end pipeline for affect estimation from noisy EEG signals, focusing on preprocessing, session-aware validation, and evaluation under limited labels.
Interactive platform for exploring endangered species data using spatial analysis and real-time filtering, designed to handle noisy and heterogeneous datasets.
Multi-step workflow system for hazard reporting and routing using NLP classification and structured orchestration, highlighting real-world input variability and system constraints.
Selected projects
Highlights from the full list: pipelines, constraints, and evaluation-first work.
EEG → model path with session-aware splits and error review before “real-time” claims.
Open in ProjectsReview-first agent for tailored résumés and cover letters: local, inspectable, human-in-the-loop.
Open in ProjectsOperational NLP under imbalance: errors drive schema and label fixes.
Open in ProjectsMedia & links