Alex Wu

I creatively build ML systems to solve unorthodox problems. Recent NYU CS + Data Science grad.

Looking for work, open to relocation

Simply put, I learn by building, parallelizing my projects to answer questions like a GPU. Looking for a bona fide team where I can apply my momentum.

Stack
Core
PythonPyTorchscikit-learnPandasSQL
ML
NLPAudio / LibrosaTime SeriesDeep Learning
Web
React / Next.jsNode.jsFastAPIGCPMongoDB
Education
2020 – 2025

New York University

B.A. Computer Science + Data Science, Minor in Mathematics

Experience
Jan 2026 – Present

Majikari

Solo Founder

Japanese collectibles price comparison across marketplaces. YC S26 applicant. Live product.

Summer 2025

Exiger LLC

Data Science Intern, NYC

Shipped NLP pipeline to production: ticket classification, priority scoring, time series forecasting with Prophet.

Summer 2023

Horizon BCBS NJ

IT Intern, Newark

Insurance catalog frontend (Salesforce/SQL), microservice monitoring, HIPAA compliance.

Research
Feb 2026 — 1st Place, Datadog Hackathon

CRUCIBLE

Two identical LLM agents play Split or Steal. No pre-programmed betrayal. They discover deception autonomously. 86% mutual destruction rate.

View Project →
NYU, Summer 2024

Audio-Based Depression Detection

w/ Pascal Wallasch

Built audio/linguistic feature pipeline for clinical depression research on 1,000+ hours of patient audio. Key finding: speech patterns vary too much across individuals for population-level detection.

2024 – Present

PsychohistoryML

ML analysis of 372 historical polities. Random Forest (AUC 0.67) finding complexity-duration relationship reverses by era.

View Project →
Projects
Audio ML

Audio Style Classifier

89.5% accuracy on 954 tracks. Handcrafted features beat CLAP/MERT by 19-24%.

CLI Tool

Suzerain

Analyze your Claude Code logs locally. See where you trust AI, where you don't.

View →
NLP

Beyond Translation

Sentence embeddings reveal translator fingerprints across 5 Nietzsche translations.

View →
Deep Learning

Music Transcription

Piano audio to MIDI. CRNN with dual-output heads for onset/duration detection.

Beyond Code
  • History: Peter Turchin's cliodynamics. Building ML pipelines around public datasets.
  • Music: Japanese indie + DnB. Why I got into audio ML.
  • Reading: Mentally somewhere in the 19th century.
  • Linguistics: How language shapes thought. Feeds into my NLP work.
  • Location: NYC/NJ, open to relocation for the right opportunity

If you're building something interesting, I'd like to hear about it.

← Back