A multidisciplinary student team connects BIM, artificial intelligence and building performance in a three-phase approach to building retrofits.
Project group F9 standing in front of their project board at the Capstone Showcase.
Students in the Capstone Design Project course partnered with Dr. Qian Chen of UBC Okanagan’s Construction Integration and Digitalization (CID) Lab. Jaeda and Colby Yurkiw were already part of the CID Lab and approached Dr. Chen about developing a Capstone project related to construction and automation.
For this Capstone team, the challenge started with a question: what if one workflow could take a building from laser scan to retrofit strategy?
The team developed a three-phase solution connecting digital building modelling, artificial intelligence and building-performance analysis.
Bringing the retrofit process together
Existing approaches typically focus on one task, such as creating BIM models, checking building-code compliance or performing energy analysis, using separate software and manual data transfer between each step. The team’s approach brought those processes together through three connected phases.
Phase 1: From LiDAR to BIM
The first phase transformed a real LiDAR point cloud into a Building Information Model (BIM), creating the digital foundation for the rest of the retrofit workflow.
Phase 2: Automating building-code compliance
The second phase introduced an AI-powered Revit plug-in to support BC Building Code compliance checks. Using a RAG-powered system built on Claude Haiku, the tool allows engineers to ask questions, evaluate the BIM model against code requirements and receive results with references to relevant code sections.
Phase 3: Evaluating retrofit options
The final phase used Pareto optimization across 122 building configurations to examine energy and carbon trade-offs and help identify potential retrofit strategies.
Together, the three phases created the workflow the students summarize simply:
“Three phases. Laser scan in, retrofit strategy out.”
The project idea originated with Jaeda after she joined the CID Lab. Through her research and two engineering internships, she experienced the time-consuming manual workflows involved in creating digital building models and verifying building-code compliance. Those experiences led her to explore how AI and Building Information Modeling could be combined.
A multidisciplinary team
The project brought together students from Civil, Mechanical and Electrical Engineering, with experience spanning construction, research, structural engineering, mechatronics and automation.
Jaeda Yurkiw and Colby Yurkiw were both research assistants with the CID Lab. Jaeda completed internships with Emil Anderson Group and ROV Engineering Consultants, while Colby completed two internships with Brookfield Global Integrated Solutions (BGIS).
Civil Engineering student Arshia Boroomandrad brought 3.5 years of construction project coordination experience to the team and is now working with Norda Stelo as a Junior Structural Engineer.
Electrical Engineering student Kurt Galanto is a registered Engineer-in-Training with APEGA, with interests in control systems, power engineering and electromechanical devices.
Mechanical Engineering student Daniel Lu concentrated in mechatronics and completed an internship with ASMPT in Hong Kong. He is interested in pursuing work in biomedical robotics.
Adapting as the project changed
One of the team’s biggest technical challenges was adapting its design as new constraints emerged throughout the eight-month project.
Phase 1 originally involved using AI to automatically segment and reconstruct point-cloud data into BIM models. Time constraints and messy, incomplete point-cloud data led the team to shift to manual segmentation. Limited datasets also meant removing cost optimization from Phase 3.
“Rather than viewing these changes as setbacks, we treated them as opportunities to refine our approach and focus on delivering an achievable solution.”
The experience reinforced the importance of being willing to adapt as new information, client feedback and real-world constraints emerged.
Supporting engineering judgment
As they developed the AI-powered tool, the students were also conscious of how it should be used in engineering practice.
Reflecting on the project, they identified intellectual honesty, communication and ethical awareness as three important qualities they developed. They were intentional about designing a tool that supports engineers in making informed decisions rather than replacing engineering judgment.
First place at the Capstone Showcase
After eight months of technical challenges, deadlines and refining their solution, the team presented the project at the UBC Okanagan Engineering Capstone Showcase and won First Place in Software and Data Systems.
The Smart Retrofit Solutions team after winning First Place in Software and Data Systems at the UBC Okanagan Engineering Capstone Showcase. From left: Daniel Lu, Jaeda Yurkiw, Colby Yurkiw, Arshia Boroomandrad, Kasra Kazerooni and Kurt Galanto.
Jaeda and Colby Yurkiw following their graduation from UBC Okanagan.
What comes next
For the team, the project’s potential extends well beyond Capstone.
“It’s exciting to think that in just an 8-month timeline, we built something with this much potential — imagine where it could be in 5 years.”
Near the end of the semester, the students met with their client specifically to explore future design opportunities and identified ways each phase could continue to develop.
Phase 1 could incorporate AI-based segmentation of point-cloud data.
Phase 2 could expand the plug-in to support building codes in provinces beyond British Columbia.
Phase 3 could introduce cost optimization, allowing users to evaluate trade-offs between different design options.
The judges saw that potential too. When the team’s award was announced, the judges noted that the solution did not really have an endpoint, which the students saw as validation of the project’s future direction.
Jaeda and Colby are already continuing that momentum by writing a research paper on their Phase 2 findings. Both are also continuing their studies in the Master of Data Science program at UBC Vancouver.