fundamental-agents 2025
Multi-agent AI platform for fundamental stock analysis: an orchestrator runs five agents to turn a ticker into an investment report.
projects
Details →
Projects
A sample of my open source projects and experiments. More on GitHub.
fisher-rao-ml 2025
An empirical study of the Fisher-Rao geodesic distance vs KL divergence in t-SNE and VAEs, with reproducible benchmarks and an arXiv-style report.
research
Details →
scaling-ensembles 2025
Do independently trained networks converge to the same function as they widen? Width sweeps measuring function-space diversity.
research
Details →
vae-playground 2025
A modular playground of five VAE variants (Vanilla, β-VAE, Conditional, VQ-VAE, WAE-MMD) with interactive marimo notebooks.
research
Details →
terror-reco 2024
Horror movie recommendations from free-text mood, via semantic search.
projects
Details →
logit-graph 2025
Fit, simulate, and compare logit graph models against ER/WS/BA via spectral GIC. On PyPI.
research
Details →
flow-match 2025
Flow Matching for generative modeling: a scikit-learn-like FlowMatcher transporting noise to data via learned ODE velocity fields.
research
Details →
kalman-transformer 2024
A Kalman filter and a Transformer as interchangeable expected-return estimators feeding a Markowitz backtest.
research
Details →
rl-rubiks-cube 2023
Self-taught Rubik's Cube solver — Autodidactic Iteration (value+policy net, no dataset) solving via beam search.
projects
Details →
bertimbau-probing 2023
Fine-tuning BERTimbau on Portuguese reviews to probe whether its embeddings recover vowel density.
research
Details →
graph-corr-embedd 2024
Measuring correlation between graphs via learned embeddings vs. classical graph-distance measures.
research
Details →