Confidential AI/RAG Systems · Backend · Operations

Augusto Oliveira

Software Engineer building confidential AI/RAG platforms, backend systems, admin control planes, and full-stack products.

I work on reliable software for complex institutional workflows where governance, auditability, and operational stability matter.

Contact Me View Selected Systems
Confidential Federal-Sector Systems Backend · AI/RAG · Data Pipelines Python · Flask · LangChain · Milvus · Docker

Confidential Systems Work

Current System

Software Engineer — Confidential AI/RAG Platform | CCSL-IFRN | Apr 2026 - Present

Confidential Federal-Sector RAG Platform

Governed AI retrieval workflows, source lifecycle control, feedback-to-knowledge promotion, auditability, and operational reliability.

I currently work on SmartFAQ, a confidential institutional AI/RAG platform used in a Brazilian federal government context. Public details are intentionally limited, but the work involves backend engineering, admin governance, vector index reliability, and operational validation.

Responsibilities

  • Built governed source lifecycle flows for controlled inclusion, update, removal, and audit of knowledge sources.
  • Improved feedback-to-knowledge workflows, promoting validated expected answers into governed RAG sources.
  • Supported hybrid index rebuild reliability across 7,434+ knowledge records while preserving governed database sources.
  • Strengthened operational visibility through status payloads, skip-reason tracing, health checks, and full-stack validation routines.

About Me

Augusto Oliveira

Software Engineer — Backend, AI/RAG Systems & Operations

I build and operate software systems that connect backend engineering, AI/RAG workflows, admin control planes, data pipelines, and infrastructure reliability.

My strongest work happens in complex institutional environments: transforming fragile processes into governed, traceable, and maintainable systems.

Based in Natal, Brazil.

GitHub Profile LinkedIn

Engineering Path

Technical Foundation

Started with HTML, CSS, and JavaScript fundamentals while building discipline in version control and problem decomposition.

Academic Consolidation

At IFRN, I strengthened software engineering fundamentals, teamwork habits, and delivery through applied projects.

Systems Expansion

Expanded into backend services, databases, and integration workflows to deliver complete end-to-end systems.

Architecture Perspective

Shifted focus from isolated features to system behavior, engineering trade-offs, and long-term maintainability.

Current Direction

Now focused on AI systems in operation, balancing product needs with reliability, traceability, and governance.

Education

March 2023 - Present

Web Development Degree

Federal Institute of Education, Science, and Technology of Rio Grande do Norte (IFRN)

Studying web technologies, software development methodologies, and digital solutions creation. Focusing on both front-end and back-end development with practical projects and industry-relevant skills.

Work Experience

Apr 2026 - Present

Software Engineer — Confidential AI/RAG Platform

CCSL-IFRN

I maintain and evolve SmartFAQ as a confidential institutional AI/RAG platform, focusing on governed source management, feedback promotion, vector index reliability, admin usability, and operational validation.

Key Achievements:

  • Built governed RAG source lifecycle flows for controlled inclusion, update, removal, and audit of knowledge sources.
  • Improved hybrid rebuild reliability across 7,434+ knowledge records while preserving governed database sources.
  • Implemented feedback-to-knowledge promotion, turning validated expected answers into governed RAG sources.
  • Improved observability through structured status payloads, skip-reason tracing, health checks, and operational validation routines.
  • Hardened the Flask-Admin experience for real operational use, improving responsiveness, wide-table handling, readability, and administrative usability.

Technologies: Python, Flask, Flask-Admin, SQLAlchemy, LangChain, Milvus, PostgreSQL, Redis, Celery, Docker Compose

May 2024 - Dec 2025

Full-Stack Developer (Extension Scholarship)

NOCS Lab - IFRN

I developed and consolidated an end-to-end vehicular access control system, integrating computer vision, real-time communication, and a web management platform, working on both development and system stabilization/validation phases.

Key Achievements:

  • Full-Stack Development: Architected and implemented front-end with Angular (TypeScript) and back-end with Node.js (Express.js), including authentication, RBAC, and REST APIs
  • Systems & IoT Integration: Implemented real-time MQTT communication between cameras, access hardware, and the central system
  • Testing & Validation: Executed manual functional tests, validated critical system flows, and verified protected route security
  • Infrastructure & Reliability: Implemented and validated automated backup system (database and images) using Docker, cron, and dedicated scripts
  • Institutional Integrations: Participated in SUAP integration using OAuth2 and JWT, extending system usage in institutional environments

Technologies: Angular, TypeScript, Node.js, Express.js, MySQL, Docker, MQTT, OAuth2, JWT

Jan 2025 - Sep 2025

Generative AI Research Scholar

CCSL-IFRN

I led SmartFAQ end to end as a functional institutional Q&A system, from data extraction and retrieval design to deployment and operational handoff.

Key Achievements:

  • RAG Architecture & Development: Designed and implemented an operational RAG architecture with Python and LangChain, using hybrid search (BM25 + Vector) and Milvus vector database for traceable and reliable responses
  • Data Engineering & Web Scraping: Developed advanced scraper with Selenium for complex government portals (Plone CMS with dynamic JavaScript), overcoming SSL errors, alerts, and async content
  • MLOps & Pipeline Orchestration: Configured and optimized ML pipelines using ZenML, Airflow, and MLflow in local and containerized environments (Docker, Docker Compose)
  • Back-end Development & UI: Created support APIs with Django and developed interactive web interface with Streamlit, deploying to Streamlit Community Cloud
  • Collaboration & Technical Management: Onboarded new scholars, collaborated with stakeholders for requirements alignment, and managed GitLab repository with branches and merge requests

Technologies: Python, LangChain, Milvus, ZenML, Airflow, MLflow, Django, Streamlit, Selenium, Docker

Engineering Focus

Backend Systems

Reliable service boundaries, auth flows, and operational backend controls.

RAG & AI Operations

Retrieval pipelines, source lifecycle rules, and auditable RAG behavior in production-adjacent use.

Full-Stack Applications

End-to-end delivery across frontend, backend APIs, auth, and operational workflows.

Data Pipelines & Scraping

Resilient extraction and processing pipelines for dynamic portals and API-ready data outputs.

Systems & Applications

Selected systems built or evolved under real operational constraints — confidentiality, reliability, integrations, data quality, and admin usability.

Selected Systems

Confidential Federal-Sector System

SmartFAQ — Confidential AI/RAG Operations Platform

Confidential institutional platform for governed knowledge operations, AI/RAG retrieval, source lifecycle control, and operational traceability.

Context

Sensitive institutional Q&A workflows with changing sources, validated feedback, and strict governance requirements.

What I worked on

  • Governed source lifecycle for inclusion, update, removal, and auditability.
  • Feedback-to-knowledge flows turning validated answers into controlled RAG sources.
  • Hybrid retrieval and rebuild reliability across governed and legacy knowledge sources.
  • Admin control plane improvements for operational visibility and safer daily use.
Institutional Access System

Vehicle Access Control System

End-to-end vehicle access control platform integrating web management, real-time events, authentication, and operational backup routines.

  • Angular dashboard and Node.js/Express backend.
  • MQTT event integration between cameras, hardware, and central services.
  • OAuth2/JWT, RBAC, protected routes, and REST API boundaries.
  • Docker-based backup routines for database and image assets.

Impact

Centralized and more reliable access operations for an institutional environment.

Stack
Angular TypeScript Node.js Express MySQL MQTT Docker OAuth2 JWT
Private Repository
Public Data Pipeline

Natal City Hall Scraper

Resilient scraping and data processing pipeline for dynamic government portal data, designed for repeatable collection and API-ready outputs.

  • Selenium extraction flow for JavaScript-rendered portal pages.
  • Handling of SSL interruptions, async content, alerts, and timing issues.
  • PostgreSQL persistence and structured data mapping.
  • Repeatable ingestion replacing fragile manual collection.

Impact

More reliable public-data collection for downstream systems and analysis.

Stack
Python Selenium PostgreSQL REST API Docker
Repository

Labs & Side Builds

Sasuke API

Sasuke Quotes API

Flask API with search, filters, and protected write routes.

Python Flask RESTful API Docker Authentication
Repository Live Demo
Jiu-Jitsu Academy Manager

Jiu-Jitsu Academy Manager

Full-stack academy management platform for enrollment, attendance, and billing.

Ruby Sinatra PostgreSQL Docker Chart.js Minitest
Repository Deploy Privado
Super Task Panel

Super Task Panel

Next.js + NestJS task workflow app with PostgreSQL persistence and OAuth.

Next.js NestJS PostgreSQL OAuth
Repository
Pokedex Ruby Project

Pokedex Ruby

Ruby/Sinatra catalog service with structured search and containerized local setup.

Ruby Sinatra SQLite Docker API Integration
Repository
Catholic Prayers Portal

Catholic Prayers Portal

Content portal with categorized navigation and maintainable backend routes.

Ruby Sinatra JavaScript CSS3
Repository
Naruto Jutsu Catalog

Naruto Jutsu Catalog

Django CRUD catalog with searchable entities and predictable data workflows.

Python Django SQLite Bootstrap
Repository

My Skills

AI Systems & RAG Operations

Retrieval Pipeline Design Source Lifecycle Workflows Incremental Indexing Strategy Hybrid Search Operations RAG Observability and Tracing LangChain + Milvus Operations

Backend Services & APIs

API Contract Design Business Rule Modeling Admin Control Workflows OAuth2, JWT, and RBAC MQTT Event Integration State Consistency Handling Flask, Django, Express Services SQLAlchemy and ORM Workflows

Data Pipelines & Automation

Dynamic Web Scraping Pipelines Async Content Handling SSL and Session Workarounds ETL and Source Parsing Routines ZenML and Airflow Orchestration MLflow Experiment Tracking API-Ready Data Shaping

Product Interfaces

Operational Admin UX Dashboard State Modeling Form and Validation Flows Angular + TypeScript React / Next.js Delivery Streamlit Support UIs Component and Layout Systems

Infrastructure & Reliability

Docker Runtime Environments Docker Compose Workflows PostgreSQL and MySQL Operations Milvus Vector Storage Operations Backup and Recovery Routines Structured Logging and Tracing Git and GitLab Delivery Flow Linux and Cron Automation

Languages

Portuguese

Native Speaker

English

Advanced

Japanese

Basic

German

Basic

Personal Notes

Watching Anime

A quiet creative influence, kept mostly as an easter egg in the interface.

Programming

Small experiments and side builds to explore tools.

Video Games

Strategy and story-driven games.

Movies

Occasional films and long-form stories.

Coffee

Part of my focused work ritual.

Beyond Engineering

Outside software, I contribute to teaching and community activities through Brazilian Jiu-Jitsu and children's catechesis.

Professional Contact

Engineering Collaboration

Open for recruiter and team conversations around backend systems, AI-enabled workflows, and full-stack engineering delivery.

LinkedIn

Augusto Oliveira

GitHub

@Augusto240

Email

jose.aug18@outlook.com
Resume / CV

Based in Natal, Brazil, currently building institutional and product-facing systems.

Uchiha Symbol