Senior Machine Learning Engineer - Semantic Spatial AI (F/M/D)
NavVis · Part-time · Posted 1w ago
When people use our technology to "bridge the gap" between the physical and digital worlds, they don’t just capture reality - they create a new one. In this new reality, they are smarter, more productive, more streamlined, and more creative - because they have the digital foundation to build the world they want to live in.
That’s what NavVis offers in all our products and services: the tools to not just map the world as it is, but to pave the way to a better future. To forge something new. Physical or digital, there is only one reality. And it’s the reality NavVis empowers people to build better.
As a Senior Machine Learning Engineer, you will identify promising advances, rigorously evaluate them on our data, and make the strongest approaches work reliably at scale. Your primary focus will be point clouds and images, and in the future may expand to Gaussian splats, 360° video, CAD, and IFC data. You will configure, train, and adapt models, then integrate successful solutions into NavVis IVION to deliver measurable customer value.
You will work with one of the world's largest collections of high-fidelity 3D point clouds and registered 360° imagery of real built environments, including construction sites, factories, manufacturing plants, and oil and gas facilities. Customers capture and upload around 500 million square meters of spatial data each year, including survey-grade point clouds with resolution as fine as 3 mm and dense registered imagery, and at petabyte scale every solution must balance quality, speed, memory use, robustness, and cost. This is a hands-on role at the intersection of computer vision, machine learning, and spatial data.
You will join the IVION Data Foundation team, a growing group of engineers and domain specialists responsible for the spatial-data backbone of NavVis IVION. The team transforms reality-capture data into a fast, streamable, high-fidelity spatial intelligence layer, spanning data ingestion, semantic enrichment of point clouds and images, algorithmic cleaning and fusion, and the data structures that enable efficient storage, streaming, and rendering.
The team works in a hybrid model, combining remote work with dedicated in-office days each week. We value spending time together in person, especially when tackling complex or ambiguous problems. We make extensive use of agentic coding tools, such as Claude Code, to amplify our impact and accelerate development. We develop on Linux or macOS, and our production systems run on Linux.
Evaluate and integrate AI/ML models, systems, and frameworks for spatial data, primarily point cloud and image data, applying them to use cases such as object segmentation, semantic scene understanding, and generating derived artifacts such as a floor layout plan
Own the practical details that determine whether a system works in production: understand its constraints, make sound architectural choices, configure it correctly, prepare and preprocess data, and optimize its performance on our datasets
Design systematic evaluations using rigorous methodology so decisions are driven by evidence rather than impressions
Diagnose, debug, and enhance ML systems when results fall short, applying a deeply analytical mindset to understand why a model behaves the way it does and how to improve it
Adapt or fine-tune existing models to our data where it delivers meaningful value, focusing on applying and getting the most from the state of the art rather than from-scratch model development and training
Work closely with your team of computer vision, computer graphics, data processing, and full-stack engineers to turn ML capabilities into valuable features our users can rely on
A Master's or PhD in a relevant field (equivalent practical experience is equally valued)
Several years of experience applying machine learning to real-world problems (exceptional candidates with less experience will also be considered)
Proven ability to adapt ML models and frameworks to real data, including data preparation, configuration, and performance optimization
A rigorous, evidence-driven approach to evaluation and experimentation, paired with strong analytical and debugging skills to understand and improve complex ML systems
Strong Python skills and experience with modern deep-learning frameworks such as PyTorch or TensorFlow (C++ experience is a plus)
A solid foundation in machine learning, linear algebra, and 3D geometry, in particular 3D coordinate systems and transformations
Experience with spatial data, especially point clouds, images, or video data (familiarity with relevant architectures such as point or vision transformers, Segment Anything models, or Grounding DINO/CLIP is a plus)
Your recruiting partner for this role is Lenka (she/her). You can expect to go through a screening call, and up to 4 rounds of interviews, where we would love to discover your passion and interests, introduce you to who we are and what drives us, and finally understand how we can potentially add value to each other's growth.
NavVis’ unwavering commitment to fostering an inclusive and diverse workplace has laid the foundation for our incredible growth. We thrive on the collective strength of our people who come from diverse backgrounds. We respect and value every experience associated with race, gender identity, sexual orientation, nationality, religion and disability. We do not discriminate on the basis of any of these, or other identities, and strongly encourage everyone to apply.
Together with you, we build NavVis!
If you need assistance at any stage of the recruiting process due to a disability, please reach out to your recruiting partner(s) for this position.