Skip to content
View HyegoJarllys's full-sized avatar

Block or report HyegoJarllys

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
HyegoJarllys/README.md

Hyego Maia

Junior AI / Data Engineer · Python · RAG & Multi-Agent Systems · Data Pipelines (GCP)


I'm a Software Engineering who already ships production-grade AI and data systems: agentic workflows, RAG assistants, and Medallion data platforms. I work at the intersection of data engineering and applied AI, building reliable pipelines, integrating LLMs with real data and tools, and turning messy processes into observable, automated systems.

  • Currently building multi-agent LLM systems and LLMOps evaluation pipelines in production.
  • Strong on data engineering (Airflow, dbt, BigQuery, Medallion architecture) and applied AI (RAG, agents, function calling).
  • Open to Junior roles in AI Engineering, Data Engineering and Data Analytics.
  • Based in Rio Grande do Norte, Brazil, working remotely with teams in São Paulo.

Featured Projects

AI Systems

Project What it is
insights-llm (Architecture Case Study) A production agentic WhatsApp analytics assistant I built (private codebase); this repo is a public demonstration of its architecture. Event-driven (FastAPI + LangChain + Gemini), read-only JWT integration to a .NET PDV, three-level authorization, outbox pattern, failure-to-defense map.
LIZ (Agentic RAG Assistant) Production-style RAG assistant with strict document grounding, async analytics and an IAM-aware architecture.
Thor (LLMOps) Autonomous LLM evaluation system. Compares prompts and models with reproducible methodology and calibrated rubrics, no external frameworks.
AI Document Extraction Pipeline End-to-end pipeline: web crawling, PDF extraction, LLM structuring, validated JSON output.

Data Engineering & Analytics

Project What it is
Olist Data Platform Enterprise-grade platform on the Olist dataset: Airflow, dbt, Docker, Great Expectations, ML drift monitoring and AI-powered diagnostics.
Data Engineering Foundations A curated foundations trail: Python/pandas, real-API ingestion, SQL analytics (joins, customer segmentation) and production-style ETL challenges in JS. Built as a deliberate skills progression.

Experience

AI Workflow Engineer, Estúdio Oggi (2026 to present) Built LLM-as-Judge evaluation infrastructure (calibrated rubrics, versioned golden sets), multi-agent squads for workflow automation, and A/B testing across Anthropic, Google and OpenAI quantifying cost, latency and quality trade-offs.

GenAI & Data Automation Intern, Maximiza IA (2026 to present) LLM-powered crawler with semantic structuring, database schema and business-rule validation, Django back-end with APIs, CRUDs and authentication.

Founder & AI Product Developer, Caatinga IA (2025 to present) Architected and deployed a municipal RAG assistant over WhatsApp; GCP data pipelines and Looker dashboards; product definition and MVP validation with stakeholders.


Tech Stack

AI & Agents: LLMs · RAG · Multi-agent orchestration · LangChain · Function calling · LLMOps (eval, monitoring, prompt/version control) Data: Python · SQL · Pandas · dbt · Airflow · Great Expectations Cloud (GCP): BigQuery · Vertex AI · Cloud Run · Pub/Sub · Looker Engineering: FastAPI · Docker · REST APIs · Event-driven architecture · MongoDB · PostgreSQL


Contact

Pinned Loading

  1. Automated-Data-Platform-Medallion-Architecture Automated-Data-Platform-Medallion-Architecture Public

    Production-grade GCP data platform with Medallion architecture, dbt, Airflow and OpenClaw (AI agent for autonomous incident response).

  2. ai-assistant-rag-liz ai-assistant-rag-liz Public

    Production-style RAG assistant for institutional use — strict document grounding, async analytics, IAM-aware architecture.

  3. ai-document-extraction-pipeline ai-document-extraction-pipeline Public

    End-to-end pipeline combining web crawling, PDF extraction and LLM-based structuring — turns unstructured documents into validated JSON.

    Python

  4. insights-llm-architecture insights-llm-architecture Public

    Architecture documentation of insights-llm, a production agentic WhatsApp analytics assistant I built (private codebase). FastAPI + LangChain + Gemini, read-only JWT integration to a .NET PDV, even…

  5. Thor-The_Judge Thor-The_Judge Public

    Thor — autonomous LLM evaluation system (LLMOps). Compares prompts/models with reproducible methodology, custom code, no external frameworks. Built for Sistema Liz municipal chatbot.

    Python

  6. data-engineering-foundations data-engineering-foundations Public

    A curated foundations trail in data engineering: Python/pandas, real-API ingestion, SQL analytics (SQLite, joins, segmentation) and production-style ETL challenges (JS). Built as a deliberate skill…

    Jupyter Notebook