Internship AI Vision Engineer

WE CANNOT PROVIDE VISA SPONSORSHIP

Details

Preferred starting date : ASAP

Weekly availability : full-time (5 days / week)

Period : 6 months

Location : Aerospace Innovation Hub, Delft, the Netherlands

About the company

Fiducial is a young but fast-growing deep-tech start-up with big ambitions at the frontier of autonomy, perception, and defence. Our core expertise lies in 3D vision and graphics technologies which we deploy in a dual-use context for inspection and defence. This role will focus mainly on the defence context where we are developing software for advanced on-board UAV situational awareness in military applications. Our solutions are aimed to operate with low-cost and widely available sensors and compute. From there, we plan to develop a line-up of interconnected solutions to safeguard European safety.

Our team consists of engineers that are passionate about the technology and solutions they develop. The only time most of us are not thinking about the technology is when we are asleep, and sometimes even then. Our team members have a background in different fields such as Aerospace Engineering, Computational Science and Engineering, Robotics and Computer Graphics. Our office is located in the Aerospace Innovation Hub, a start-up hub located on the TU Delft campus in the faculty of Aerospace Engineering.

We work closely with top-tier partners, ranging from government agencies and prime contractors to academic research institutes and other start-ups. Our projects span from large tender orders in collaboration with partners to low TRL research with government agencies. Whether through formal R&D programs or rapid prototyping tracks, we operate at the intersection of innovation and deployment.

Your role

You’ll be responsible for adapting, training, and testing object detection algorithms for several of our applications. You’ll work closely together with the rest of the engineering team to integrate your networks and algorithms with other parts of our solutions and test them in the field.

What we expect from you

• Familiarity with commodity ML concepts and frameworks.
• Ability to quickly spin up commodity training runs such as object detection, instance segmentation etc in a few hours.
• High familiarity with TensorRT, ability to convert custom models to ONNX.
• High speed in understanding, setting up, running and learning from the most arcane open-source repositories.
• Intuition for debugging your networks. Ability to understand why a given network learned something and not another thing given a dataset and architecture.
• Big plusses :
• Intuition for learnability of uncommon target distributions, mainly related to vision concepts.
• Experience with large(r)-scale training runs.
• Full time availability (40 hours)

While having the perfect background is nice, if you are able to present evidence of outstanding engineering and execution capability we are interested in talking regardless. If you are the right fit, we believe in your ability to pick up the required background knowledge on the fly.

What we offer

The application procedure