Mateusz Dalba
I build production AI systems—from computer vision and predictive models to LLM applications and scalable data platforms.
I’m a Senior Machine Learning Engineer and AI Consultant based in Poland. I help teams turn business problems into production AI systems.
My work spans computer vision, predictive modeling, LLM applications, and data platforms across healthcare, manufacturing, retail, and other industries. I work across the full delivery lifecycle—from understanding requirements and preparing data to developing models, integrating APIs, and deploying solutions.
My core toolkit includes Python, Databricks, PySpark, Azure, FastAPI, and modern machine learning frameworks. Alongside engineering, I bring experience in client-facing consulting, technical mentoring, and teaching.
MSc in Advanced Analytics – Big Data and a bachelor’s degree in Quantitative Methods and Information Systems from the Warsaw School of Economics. My master’s thesis explored backdoor attacks against deep learning models.
2025.06 – present
SoftServe · B2B contract
Delivering machine learning, data engineering, and computer vision solutions for marketing analytics and healthcare.
2024.02 - 2026.05
Developing computer vision, optimization, LLM applications, and data platforms for industrial and sports technology clients.
2022.03 - 2024.06
As a member of the AI Engineering team, I have implemented and consulted on various advanced analytics projects across different industries.
2023.05 - 2023.12
Designed and implemented a CV module from scratch in inteligent shopping carts product.
2022.05 - 2025.05
As a mentor at Future Collars, I taught multiple groups of learners through comprehensive courses on Python, SQL, and Machine Learning.
2020.08 - 2022.03
Data Scientist in Research and Development team working on various AI/ML related projects. This was my first job as a Data Scientist which started as non-paid internship.
2018.05 - 2018.12
Member of Statistical Programming Team. This was my first job in data industry.
Selected work
Selected commercial engagements across machine learning, computer vision, and generative AI. Client names and implementation details are omitted. Images are AI-generated concept illustrations, not client screenshots.
01 / Marketing analytics
Developing data pipelines and machine learning models for commercial attribution and media mix modeling.
My work includes scalable data processing, regression and CatBoost models, data drift monitoring, and validation workflows. I collaborate with business and engineering stakeholders to translate analytical questions into implementable solutions.
02 / Healthcare
Developing visual analysis software for monitoring medical staff preparing for surgical procedures.
My contribution covers human pose estimation, 2D/3D visual analysis, object detection, tracking, and activity analysis, alongside integration of AI components into Python backend services.
03 / Generative AI
Building evaluation workflows to assess and improve LLM-based agents across test scenarios.
I designed and implemented an agent evaluation pipeline using Langfuse, OpenAI, and Azure ML Studio. The engagement also included microservices integrating text-to-speech and speech-to-text capabilities.
04 / Manufacturing
Developing glass inspection algorithms, production optimization, and analytics for industrial installations.
My work included detecting glass defects and cracks, optimizing glass placement and loading-table sorting, and developing an analytics platform for processing data from hundreds of client installations worldwide. The data stack included Azure, PySpark, and Superset.
05 / Technical support
Developing an LLM-based assistant for technical inquiries using existing support documentation.
The solution brought together technical instructions, support knowledge, and documentation to assist with customer inquiries. My work included LLM integration and application development using OpenAI, Azure, and Django.
06 / Retail
Developing a computer vision algorithm powering product recognition in an intelligent shopping cart.
I developed a bespoke computer vision algorithm for the cart’s core functionality. The work involved product recognition and tracking in a dynamic retail environment, with varying lighting conditions and limited camera perspectives.
Further work across forecasting, computer vision, analytics, and developer tooling.
A forecasting application integrated into a demand planning workflow, combining time-series models with interactive analytics.
Time Series · Python · Azure
Tools for automating repetitive analytical tasks and bringing notebook-based methodologies into reusable applications.
Scikit-learn · Streamlit · Django
Computer vision for recognizing Polish banknotes, designed to support people with visual impairments.
Computer Vision · TensorFlow Lite · OpenCV
An application for analyzing telemetry from simulators and real cars to support driving-style analysis.
Telemetry · Machine Learning · Plotly
Exploring virtual clothing try-on using image segmentation, generative models, and human keypoint detection.
Generative Models · Segmentation · TensorFlow
Sentiment analysis, keyword extraction, and semantic similarity applied to documents, support tickets, and customer feedback.
NLP · Hugging Face · PySparkLooking for support with an AI project or an experienced ML engineer for your team? Let’s talk.