Hey!
I am Konstantinos Kyrtsonis!
Passionate about developing innovative solutions in fast-paced environments, with over 4 years of experience in the data field I am a mission-oriented professional with expertise in AI Engineering, LLMs, Computer Vision and Data Science.
What I love doing: being curious about analyzing data, creating automated pipelines, finding cause and effect, and explaining them to people. Finding the optimal way to use LLMs and RAG to solve business problems is what drives me. I am interested in finding ways to make innovative solutions not only as research but to solve real business problems.
Feel free to get in touch or take a look at my past work below.
Portfolio
A selection of AI engineering, data science, and construction-tech projects.
AI & Data Science
Machine learning, NLP, computer vision, and data-driven solutions for industry
BIM–Schedule Integrated Pipeline & Classification
AI-powered construction data classification and 4D simulation pipeline
Description
Architected a system to classify BIM objects and schedules by collecting, cleaning, and processing client input data through a multilayer pipeline powered by advanced LLM-based classifiers fine-tuned for construction data—mapping objects to tasks for analytics and prognoses.
Details
Designed and built an end-to-end AI platform that ingests client BIM models and construction schedules, classifies IFC building components using fine-tuned large language models, and automatically maps them to schedule tasks for 4D simulations and logistics planning. The system incorporates unique object hashing to eliminate redundant BIM classification, rule-based and ML-based labeling of construction elements, efficient Parquet-based data handling, and optimized mapping between IFC components and schedule activities. A performance optimization initiative reduced processing times from days to under 5 minutes for typical multi-storey buildings.
References & Technologies
IFC (Industry Foundation Classes) · buildingSMART · Large Language Models · Construction 4D Simulation
Solar panel efficiency
Description
This is the final project during my 3 month bootcamp studies at Schaffhausen Institute of Technology. It was a collaboration with Nispera AG Zurich.
Point Cloud Semantic Segmentation
Description
Point Cloud Classification with PointNet. Input of dataset with point clouds of different objects. Creation of the PointNet architecture from the bottom, train it on the dataset, and then do predictions on new point clouds that the neural network has not seen before.
Concrete Crack Detection
Description
This repository contains the code for crack detection in concrete surfaces. It is a PyTorch implementation of the paper by Young-Jin Cha and Wooram Choi Deep Learning-Based Crack Damage Detection Using Convolutional Neural Networks
Predicting Home Prices
Description
This data science project series is about building a real estate price prediction website. First, I built a model using sklearn and linear regression using Melbourne home prices dataset from kaggle.com.
Construction Audio Event Classification
Sensor-based monitoring and audio event detection for construction logistics
Description
Designed and proposed a sensor-based monitoring pilot for construction logistics, using audio-based event detection to measure truck arrivals, congestion, unloading duration, material-handling intensity, and site bottlenecks.
Details
Developed an audio event classification system targeting real-world construction site environments. The pilot leverages microphone sensor arrays and deep learning models to detect and classify logistical events—truck arrivals and departures, concrete pouring, crane operations, and material unloading—in real time. Extracted features feed into dashboards that quantify congestion, unloading duration, and material-handling intensity, enabling data-driven decisions for construction logistics and last-mile management workflows.
References & Technologies
Audio event detection · Construction logistics · Last-mile management · Sensor-based monitoring · Deep learning for audio
Construction Schedule Analytics & Insights
Automated schedule analysis, peak-phase detection, and risk identification
Description
Created a pipeline system that reads client construction schedules and provides actionable insights on peak phases, resource bottlenecks, critical-path risks, and potential scheduling conflicts.
Details
Built an automated analytics pipeline that ingests construction project schedules in various formats, normalizes and parses activity data, and produces comprehensive reports highlighting peak construction phases, resource allocation imbalances, critical-path vulnerabilities, float erosion risks, and potential cascading delays. The system provides visual dashboards and early-warning indicators to support proactive decision-making by project managers and planners.
References & Technologies
Critical Path Method (CPM) · Construction scheduling · Resource leveling · Risk analysis · Project controls
Interactive BIM Viewer
Explore a 3D building model with professional tools — select objects to inspect properties, clip sections, switch rendering modes, or walk through the building in first person.
Loading BIM model…
Click any element to view its IFC properties. Hover to highlight. Double-click in clipping mode to place section planes.
Skills
Technologies & Tools
Through hands-on engineering and continuous learning, I have practical experience with:
Applied to domains such as:
Experience & Qualifications
I combine academic depth with industry practice—studying Data Science, Technology and Innovation at the University of Edinburgh (Top 15 worldwide in 2023) while working full-time as an AI Engineer:
- Currently at: Amberg Loglay AG, Switzerland
- Data Science, Technology and Innovation MSc — University of Edinburgh
- Data Science Bootcamp — Constructor University
Contact
Have work opportunities? Want to collaborate on a project? Need advice? Or maybe grab some coffee?
You can shoot an email to kyrkost@outlook.com