AI Engineer/Computer Vision

Ambush; Porto Alegre; On-site; BRL 180,000 - 300,000; Full time

Job summary

Ambush is building a next-generation real estate investment platform that replaces manual processes and scattered PDFs with reliable data and automated workflows. You will teach computers to read complex documents and convert unstructured data into structured signals that power our matching engine.

In this role, you will enhance geometry extraction from floor plans using Evision and OCR, build scalable LLM pipelines for data extraction, and collaborate to feed our recommendation funnel with

Qualifications

• Solid, hands-on experience with Computer Vision and OCR.
• Real-world experience with RAG (Retrieval-Augmented Generation) and using LLMs for data extraction.
• Proven ability to solve complex problems while dealing with highly unstructured data.

Responsibilities

• Actively improve the geometry extraction from floor plans and land plots using Evision (OpenCV for Elixir) combined with OCR.
• Build and scale LLM solutions to extract and structure information from diverse documents (e.g., inclusion lists, builder profiles, and facades).
• Collaborate to ensure this extracted intelligence feeds our recommendation funnel, connecting visual data with tools like PostGIS and vector databases.

Skills

Computer Vision

OCR

RAG

LLM data extraction

Unstructured data

Tools

Elixir

PostGIS

Vector databases

Embeddings

Job description

We are building a next-generation real estate investment platform, replacing manual processes and scattered PDFs with reliable data and intelligent automation. Think of us as an automated matchmaker: we pair land plots with compatible house designs, optimizing for budgets, zoning rules, and investor goals at scale.

In this role, you will teach computers to "read" complex visual and textual documents and respond based on them. Because our source data arrives entirely unstructured, from floor plans to inclusion lists, your core challenge is transforming that into the structured data driving our matching pipeline.

When you join us, you will:

• Actively improve the geometry extraction from floor plans and land plots using Evision (OpenCV for Elixir) combined with OCR.
• Build and scale LLM solutions to extract and structure information from diverse documents (e.g., inclusion lists, builder profiles, and facades).
• Collaborate to ensure this extracted intelligence feeds our recommendation funnel, connecting visual data with tools like PostGIS and vector databases.

MUST-HAVES

• Solid, hands-on experience with Computer Vision and OCR.
• Real-world experience with RAG (Retrieval-Augmented Generation) and using LLMs for data extraction.
• Proven ability to solve complex problems while dealing with highly unstructured data.

(Nice-to-haves: Familiarity with Machine Learning, Elixir, vector databases, embeddings, or PostGIS is a big plus!)

Ambush is a people first company where talented, thoughtful individuals come together to build meaningful products and lasting partnerships. We believe the best work happens when people feel supported, trusted and empowered to bring their full abilities to the table. Since 2015, because of our people first and long term mindset, we have grown into a partner relied on by some of the best companies in the world. We combine strong engineering, design and strategy with a growing strength in AI to help our clients see what is next and achieve bigger outcomes. At the heart of everything we do is our team. We collaborate, take risks, lift each other up and take pride in doing work the right way, not settling for a quick makeshift solution. If you join Ambush, you join a group of people who want you to succeed and who show up for each other every day. We believe in and expect real teamwork, a constant drive to be better, and delivering meaningful long term outcomes that we can be proud of together.