Senior Computer Vision Engineer

Munich Germany Ref: 2020

About Our Client

Our client is developing a no?code, agent?driven AI platform tailored for the Architecture, Engineering, Construction (AEC) sector and other physical?world industries. Their technology focuses on automating complex, expertise?heavy workflows that traditionally rely on fragmented systems, undocumented specialist knowledge, and isolated data sources.

Their multi?agent framework enables advanced reasoning across real?world environments and converts previously bespoke, service?intensive processes into scalable, software?driven automation.

They are relied on by several major AEC organisations across Europe, North America, Latin America, and Asia.

The Role

In this position, you will take the lead on designing and advancing the spatial intelligence engine that powers our client’s agentic AI platform. The work sits at the convergence of 3D reasoning, geometric machine learning, and agent?based inference. You’ll be responsible for building end?to?end systems that turn raw, multi?modal inputs into structured, machine?readable spatial knowledge.

Your priority will be accurate topology, semantic fidelity, and engineering-grounded understanding rather than photorealistic rendering. You will architect the computational logic enabling autonomous agents to interpret, navigate, and reason about 3D environments – ensuring insights and decisions are rooted in real?world physical constraints.

Responsibilities

Spatial Reconstruction & Semantics

• Develop methods for translating real?world spaces into geometric and semantic 3D representations, including occupancy grids, volumetric segmentation, and primitive?based reconstruction.

Multi?Modal Integration

• Create fusion pipelines that combine foundational model outputs (e.g., segmentation, panoptic labels) with 3D geometry to produce high?quality, structured scene representations.

Knowledge Graph Enrichment

• Automate the transformation of reconstructed 3D environments into spatio?temporal knowledge graph structures, capturing functional hierarchies and relationships between built elements.

Agentic Reasoning

• Build inference workflows that enable agents to perform high?level reasoning directly on spatial knowledge graphs.

Affordance Discovery

• Implement models that identify functional possibilities (“affordances”) within a 3D environment based on geometry, constraints, and engineering logic.

Performance & Scale

• Bring state?of?the?art research into production environments and ensure reliability, scalability, and measurable value across customer deployments.

Requirements

• MS/PhD in Computer Science, Robotics, Electrical Engineering, or a similar discipline
• 3+ years of hands?on experience in computer vision or deep learning
• 2+ years leading 3D computer vision initiatives, including geometric ML, reconstruction, or related domains
• Experience with physics simulation tools (e.g., Isaac Gym, MuJoCo, PyBullet) is advantageous
• Familiarity with agent?based AI frameworks (e.g., GraphRAG, LangGraph, LlamaIndex) is beneficial
• Strong implementation skills with Open3D, PyTorch3D, multi?view stereo, TSDF, and 2D?to?3D lifting
• Solid understanding of scalable software design
• Previous exposure to startup?style, fast?paced environments is a plus
• Fluent English communication

Why Work With Our Client

• Innovation at the Edge : Help define the next generation of AI for spatially complex, real?world industries
• High Ownership : Shape how multi?modal reasoning and 3D scene understanding operate at scale
• Exceptional Team : Collaborate with global experts across ML, software engineering, and AEC domains
• Strong Growth Potential : Competitive package with meaningful long?term upside