Skip to content
View LukeSantossz's full-sized avatar
👽
Always in a git rebase --interactive with life.
👽
Always in a git rebase --interactive with life.
21 followers · 28 following

Achievements

Achievement: Pull Sharkx3Achievement: Pair Extraordinairex2Achievement: QuickdrawAchievement: YOLO

Achievements

Achievement: Pull Sharkx3Achievement: Pair Extraordinairex2Achievement: QuickdrawAchievement: YOLO

Highlights

  • Pro

Block or report LukeSantossz

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
LukeSantossz/README.md

AI Engineer building RAG and agent systems. I work on the part after the model works: grounding, verification, and the limits that keep it honest.

Open to remote roles, international or Brazil-based · GMT-3

Open source

What it is Proof Stack
sb100_agents Self-hostable RAG API that scores its own confidence Agent gate calibrated to 96.7% / 3.3% TPR-FPR · 372 tests, 89.8% coverage · 15 ADRs Python FastAPI Qdrant Ollama LangGraph
tweet-sentiment Sentiment classification fine-tuned end to end 0.887 macro F1, +51.9% over the frozen-features baseline · 28.5× faster Rust preprocessing, parity validated at 3 scales Python Rust PyTorch Polars
weather-forecast scikit-learn model running in the browser Matched to 1e-6 against Python · 200-tree forest serialised and traversed in TypeScript · test-first across 20+ features, traceable in the log Python TypeScript LightGBM
visiosoil-app Soil texture from a photo, fully offline 3rd of 1,300+ at FETEPS 2025 · paper published at ICPA/ConBAP 2026 · on-device inference in a killable isolate, group-aware dataset splits Flutter Dart TFLite

Built at work

Closed source. Numbers verified against the repositories I contributed to.

What it is Proof Stack
Agent code-generation platform Spec in, working application out 103k lines, 3,397 tests · generated code isolated in per-job micro-VMs · deterministic security gate Python LangGraph Kata Docker
AI advising platform Grounded answers over institutional data Hybrid retrieval built from scratch over pgvector · refuse-before-call gate and post-answer verifier · 854 tests Python FastAPI pgvector Anthropic

My AI Engineering Stack

Agents & Orchestration

RAG & Retrieval

ML & Inference

Backend & Data

Infra & Quality

GitHub Activity

Pinned Loading

  1. tweet-sentiment-analysis tweet-sentiment-analysis Public

    A Natural Language Processing (NLP) project focused on Exploratory Data Analysis (EDA) and sentiment classification of Twitter data using Machine Learning.

    Python

  2. visiosoil-app visiosoil-app Public

    Mobile app built with Flutter and Dart, integrating Machine Learning and Computer Vision for real-time soil texture classification

    Dart

  3. sb100_agents sb100_agents Public

    RAG-based agricultural Q&A system using FastAPI, Qdrant vector database, and Ollama LLM for semantic document retrieval and response generation from PDF knowledge bases.

    Python 1 3

  4. weather-forecast weather-forecast Public

    End-to-end temperature forecasting pipeline: 0.27°C RMSE across 211 countries. Gradient boosting (sklearn/LightGBM) and statistical baselines (ARIMA/SARIMA/Prophet) in an inverse-RMSE weighted ense…

    Python