Skip to content
View rafael-robsonn's full-sized avatar
1 follower · 3 following
  • Brasilia
  • 12:46 (UTC -03:00)

Highlights

  • Pro

Block or report rafael-robsonn

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.

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
rafael-robsonn/README.md

Rafael Robson Nunes de Araújo

Typing SVG


LinkedIn  Email  GitHub


Profile Views


> whoami

role   : CEO, Chesin
focus  : AI-native data engineering
base   : Brasília, DF — Brazil
study  : Data Science & ML @ CEUB + USP
status : building

I founded Chesin with two bets one automates data pipelines with AI, the other reinvents urban mobility with intelligent electric fleets.

No demos. Real systems.

> ls ./now

currently = {
  "shipping" : "DataSM v1 — Decision Layer",
  "target"   : "≥ 85% code acceptance rate",
  "reading"  : "Designing ML Systems",
  "open_to"  : [
    "data engineering roles",
    "AI/ML engineering",
    "founding team"
  ],
}

◈ Featured Work


⬡ DataSM   [flagship]

AI agent that writes, reviews and deploys production data pipelines without a human in the loop.

Built on a dual-brain architecture: LangGraph orchestrates the full pipeline while Claude acts as an adversarial Reviewer, catching hallucinations and enforcing schema correctness before anything touches production. A Decision Layer (replacing the broken weighted-blend merge) arbitrates between agent outputs. dbt + BigQuery as the initial wedge.

North star metric: ≥ 85% code acceptance rate.

User intent
    │
    ▼
[LangGraph Orchestrator]
    ├── Codestral (Generator)
    ├── Claude (Reviewer / Critic)
    └── Decision Layer
            │
            ▼
    [dbt → BigQuery] ── Production

Python LangGraph Claude React dbt BigQuery

⬡ ChesinBus

Intelligent fleet management for hybrid and electric buses.

Real-time telemetry ingestion, route optimization, and predictive maintenance at scale. Built for the operational complexity of emerging markets where data is sparse and downtime is expensive.


Python Spark PostgreSQL IoT



⬡ gemma-cli

Rust CLI with full streaming, runtime flags and 4 subcommands (ask chat models info) integrating Gemma 4 E4B via Ollama locally.

Rust Ollama
↳ more projects
Project Description Stack
Language Quest Gamified language learning — progression system, adaptive quizzes, real-time feedback React, TypeScript
Java Forum App Full MVC application — auth, threads, points system, end-to-end tests Java, Spring, PostgreSQL, DBUnit, Selenium
DataSM Code Review Automated code review dashboard — dual-agent pipeline (Reviewer + Critic), 5-stage WebSocket flow React, Python, WebSocket
nlp-regras-basico Study notes: BoW, TF-IDF, Word2Vec, GloVe, RNN, LSTM, GRU + NLPortugues/USP labs Python, PyTorch

◈ Hackathons


Transforming Enterprise Through AI

Tessera

"The trust layer for enterprise AI agents."

Real-time multi-agent threat chain detection with LGPD compliance enforcement and cryptographically-signed audit trails. Policy engine with regulatory basis, DPO settings, RoPA (Art. 37), health checks, and a SOC-grade dark UI built for long analyst sessions. Threat severity classification with audit blockchain verification.


React Python LGPD Crypto

AMD Developer Hackathon

TokenBurners

"We burn tokens so you don't have to."

Automated code review system powered by adversarial AI. Dual-agent pipeline (Reviewer + Critic) in 5 stages with real-time feedback via WebSocket. Issues classified by severity critical, major, minor, suggestion — with per-file stats and tree visualization. AMD ROCm dark mode UI.


React Python WebSocket AMD

◈ Stack


Skills

Skills

Skills


↳ full breakdown
Layer Technologies
AI / LLM LangChain, LangGraph, RAG, LSTM/GRU, LLM Fine-tuning, Codestral, Claude API
Data Engineering Apache Spark, dbt, BigQuery, PostgreSQL, SQL, ETL/ELT
Languages Python, Scala, Java, Rust, TypeScript, COBOL
Backend FastAPI, Spring Boot, Node.js, REST, MVC
Frontend Javascript, React, TypeScript, HTML/CSS
BI Power BI
CLOUD AWS, GCP, AZURE
Infra Docker, Git, Maven, Tomcat, DBUnit, Selenium, Terraform

◈ Trophies


trophy


◈ Stats





systems that work when no one's watching



Pinned Loading

  1. nlp-regras-basico nlp-regras-basico Public

    Notas de estudo sobre NLP semântica básica, modelos probabilísticos, redes neurais, vetorização de texto (BoW, TF-IDF, Word2Vec, GloVe) e arquiteturas recorrentes (RNN, LSTM, GRU). Material de estudo.

    1