High Level Description
Modern workplaces increasingly aim for intelligent, responsive environments. One compelling application is automated people counting in meeting rooms, tied to smart lighting control. Rather than always relying on cloud-based inference, this thesis investigates how data pipeline strategies and edge–cloud coordination can make distilled, compact on-device models viable for real-time decisions, such as balancing latency, bandwidth, energy, and accuracy.

Project Description
You will design and implement a people-counting and lighting control system for meeting rooms. A camera or depth sensor plus a micro-computer device (e.g. Raspberry Pi) will capture data; the system decides whether lights should be based on occupancy and ambient illumination. Key work will focus on data pipeline architecture, such as frame sampling, compression, caching, and routing decisions, and how those pipeline strategies impact whether the edge model is selected over a cloud-based teacher for inference.

The thesis work may include:

• Building a small people-counting student model and a stronger cloud teacher model for comparison.

• Comparing edge inference vs. cloud inference under various pipeline configurations.

• Experimenting with preprocessing, adaptive sampling, serialization formats, caching strategies, and routing logic.

• Quantifying trade-offs: latency, bandwidth, accuracy, and energy consumption.

• Validating on real meeting room scenes in office conditions.

Who are we looking for?
We're searching for master's or bachelor students who are passionate about AI, IoT, and smart systems, and who enjoy building solutions that connect the physical and digital worlds. Experience with Python or basic computer vision is a plus, but not necessary. If you’re comfortable with any programming language, you’ll quickly pick up the tools used in this project.

You should also:

• Enjoy solving real-world problems and experimenting with new ideas.

• Be motivated to learn how AI and data pipelines can work together to make intelligent systems faster and more efficient.

• Be fluent in Swedish, written and spoken.

Purpose
We’re exploring how AI at the edge can deliver real-time insights without always depending on the cloud. The plan is to combine lightweight, distilled models with smart data pipeline design to make people-counting systems both fast and efficient. By balancing local and cloud processing, we aim to show how edge AI can be practical, responsive, and energy-aware in real office environments.

An Exciting Journey with Knightec Group
Semcon and Knightec have joined forces as Knightec Group. Together, we are Northern Europe’s leading strategic partner in product and digital service development. With a unique combination of cross-functional expertise and a holistic business understanding, we help our clients realize their strategies – from idea to complete solution.

Practical Information
This is a thesis position, located at our office in Gothenburg. Start date 2026-01-20.

Please submit your application as soon as possible, but no later than 2025-11-30. If you have any questions, you are welcome to contact Felix Fredin. Note that due to GDPR, we only accept applications through our careers page.