Skip to content

Navigation Menu

Sign in
Sign up
@garuyto09
garuyto09
Follow

Gabriel Henrique Cerqueira garuyto09

Block or report garuyto09

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
garuyto09 /README.md

Gabriel Henrique Cerqueira

Applied Mathematician · Credit Risk Analyst @ Mercado Pago · Brazil

Applied mathematician (B.Sc., UNICAMP) turning large-scale financial data into decisions. I work on the models behind consumer credit — who gets credit, how much, at what price — which in practice means sitting between risk modeling, large-scale SQL, and translating portfolio behavior into something a business can act on.

Outside of work the interest is the same one from a different angle: what a model can actually prove, and what it only appears to. Usually that means probability, game theory, and simulations that turn out to be measuring the wrong thing.


What I work on

Credit risk modeling — PD scorecards, expected loss, portfolio segmentation, policy design across risk tiers

Machine learning — classification benchmarks, feature importance, calibration and the gap between ranking well and being right about the level

Applied math & game theory — agent-based simulation, equilibrium and strategy analysis (Truco), gradient descent, simplex

Statistics & experimentation — A/B testing, cohort · vintage · funnel analysis, statistical power, and the difference between significant and worth anything

Data engineering & visualization — BigQuery at scale, cost-aware SQL, ETL, dashboards and metrics layers people actually steer on

Stack

Python Julia R SQL SAS

pandas NumPy scikit-learn seaborn Matplotlib tidyverse

BigQuery Looker Docker Git Jupyter

Selected work

credit-default-models · Python scikit-learn Benchmark of classification models for credit default prediction, evaluated on KS, Gini, PSI and calibration rather than accuracy. Undergraduate research recognized among the 100 best projects at UNICAMP.

truco-julia · Julia Agent-based study of Truco — 30,000 agents, 100,000 matches. Card strategy turns out to matter fifteen times less than aggression, and the original notebook could not see it because it was measuring at the wrong level.

Languages

Portuguese English Spanish Mandarin German French Japanese

I document in English and present credit policy in Spanish and Portuguese.


📫 ghcerq@gmail.com · open to Senior Data Analyst / Data Visualization roles

Popular repositories Loading

  1. credit-default-models credit-default-models Public

    Credit default prediction benchmark evaluated on KS, Gini, PSI and calibration — not accuracy. UNICAMP undergraduate research.

    Python

  2. truco-julia truco-julia Public

    Agent-based study of Truco in Julia: 30k agents, 100k matches — aggression beats card strategy 15 to 1

    Jupyter Notebook

  3. garuyto09 garuyto09 Public

    Profile README

AltStyle によって変換されたページ (->オリジナル) /