> EMBEDDING documents_

Data that becomes intelligence that becomes product.

I'm Helena Gonçalves Schwantes (Lena), a Data Engineer & LLM Integration Specialist. I've pushed 100M+ records a day through dbt and Snowflake, and now I build GraphRAG and multi-agent systems that leave the notebook and actually ship.

Get in touch See my work
Scroll
100M+records a day through my dbt + Snowflake pipelines
2+ yrsof enterprise data engineering at EY
4AI and data products built end to end

What I do

01 / 03

Data

Pipelines that hold up at scale, from dbt models over 100M+ daily records to vectorization flows that turn messy documents into clean, searchable data.

  • Python
  • Pandas
  • dbt
  • Snowflake
02 / 03

Intelligence

GraphRAG, knowledge graphs and multi-agent systems, including local LLMs for sensitive data that cannot leave the company.

  • LangChain
  • GraphRAG
  • Weaviate
  • FalkorDB
  • Prompt engineering
03 / 03

Product

Shipping it: SaaS, automations and recommendation agents that real users touch every day.

  • FastAPI
  • SaaS
  • Agents
  • Automation

Selected work

Vectorization pipeline plus GraphRAG over large document sets, combining embeddings with a knowledge graph for grounded answers.

  • tiktoken
  • Voyage AI
  • Weaviate
  • FalkorDB

Fiscal audit with local multi-agent AI, cross-checking NF-e and DANFE against purchase orders, e-mails and ERP records.

  • Local LLMs
  • Multi-agent
  • NF-e
  • ERP

Automation of affiliate offers from Shopee, Mercado Livre and Amazon, published to WhatsApp groups.

  • SaaS
  • Automation
  • WhatsApp

A curated collection of trending Python and JavaScript repos from 2023 to 2026, with a recommendation agent on top.

  • Python
  • JavaScript
  • Agents

Experience

  1. Now

    Applied AI R&D Fundação CERTI

    Applied research projects that turn LLMs into working systems: a GraphRAG and vectorization pipeline over large document sets, and a multi-agent fiscal audit running on local LLMs, so sensitive data never leaves the building.

  2. Aug 2022 to Nov 2024

    Data Engineer EY Consulting

    Owned large-scale dbt + Snowflake pipelines processing 100M+ records a day. Contributed to the data architecture of a transactional financial data project.

  3. Before tech

    Early Childhood Teacher

    Where I learned to explain complex things simply, and to stay patient with systems and people that do not behave as documented.

  4. In progress

    Technologist in Big Data & Analytical Intelligence UNISUL

About

Systems need to be useful, not just technically correct.

I started my career teaching kindergarten, which turned out to be solid training for data engineering: both involve unpredictable inputs, a lot of patience and explaining the same thing in three different ways until it clicks. Then I moved to tech and spent over two years at EY keeping 100M+ records a day flowing through dbt and Snowflake.

Today I build applied AI: GraphRAG, vector pipelines and multi-agent systems on local LLMs, plus my own products on the side. I'm looking for Senior Python, LLM and AI Specialist roles where shipping something useful matters more than the buzzwords around it.

Based in
Florianópolis, Brazil
Looking for
Senior Python, LLM and AI Specialist roles
Studying
Big Data & Analytical Intelligence, UNISUL
PythonFastAPILangChainPandasdbtSnowflakeWeaviateFalkorDBRAGGraphRAGPrompt engineering

Let's build something intelligent.

helena@schwantes.com.br