┌─ whoami ───────────────────────────────────────────────────────────────┐
│ │
│ AI Engineer. I work where models meet systems. │
│ │
│ Computer vision, tabular ML, data pipelines and retrieval-grounded │
│ LLM workflows, shipped behind APIs and interfaces people actually use.│
│ │
│ Most of it is quiet work: data you can trust, evaluation you can │
│ defend, deployments that stay up. │
│ │
└────────────────────────────────────────────────────────────────────────┘
┌─ stack ────────────────────────────────────────────────────────────────┐
│ │
│ . │
│ ├── learn/ python · pytorch · tensorflow · scikit-learn · xgboost │
│ ├── see/ opencv · yolov8 │
│ ├── store/ postgresql · sqlite · etl pipelines │
│ ├── serve/ fastapi · docker · react · next.js · streamlit │
│ ├── ground/ rag · langchain · faiss · groq · gemini · ollama │
│ └── ship/ vercel · hugging face · gcp · aws · firebase │
│ │
└────────────────────────────────────────────────────────────────────────┘
01 |
DocMind | grounded Q&A over documents, every answer tied to its source | src |
02 |
Industrial Anomaly Monitor | autoencoder watching a replayed sensor stream | src |
03 |
PlatePulse | two-stage ALPR for Brazilian plates, YOLOv8 then OCR | src |
04 |
Corporate Signal | anomalous company events, turned into executive briefings | src |
05 |
PM Monitor | rolling LSTM, multivariate sensors into failure probability | src |
06 |
Visual Anomaly Lab | denoising autoencoder with heatmaps that explain the defect | src |
07 |
CineScope | semantic film search by theme and mood | src |
08 |
Gray Matter LABS | research agent over arXiv, web and Wikipedia | src |
09 |
RL Portfolio Ops | PPO allocation policies under risk guardrails | src |
Nine live URLs, zero notebooks. Name opens the demo, src opens the repository. Write-ups at sidnei-almeida.github.io/projects.
┌─ method ───────────────────────────────────────────────────────────────┐
│ │
│ 01 frame outcomes, latency budgets, drift risk, labeling cost │
│ 02 land schemas, leakage, reconciliation, repeatable ingestion│
│ 03 transform features that match train and inference payloads │
│ 04 evaluate baselines first, metrics aligned with failure modes │
│ 05 wire thin services, contracts that are easy to test │
│ 06 ship deploy, observe, patch data, refresh models │
│ │
└────────────────────────────────────────────────────────────────────────┘
┌─ contact ──────────────────────────────────────────────────────────────┐
│ │
│ portfolio sidnei-almeida.github.io │
│ linkedin linkedin.com/in/saaelmeida93 │
│ email [email protected] │
│ │
│ status open to AI / ML engineering work │
│ │
└────────────────────────────────────────────────────────────────────────┘