resume profile picture
Sigurd Roll SolbergData Scientist / ML Engineer
Email
[email protected]
Phone
+47 46822072
Location
Brussels, Belgium
20/07/1999
Norwegian
LinkedIn
linkedin.com/in/sigurd-solberg
GitHub
github.com/SigurdSolberg
https://www.sigurdsolberg.com/
Profile

Data Scientist and Machine Learning Engineer with strong foundations in mathematics and modern deep learning, and hands-on experience building, deploying, and operating large-scale AI systems in regulated production environments.

I specialize in representation learning, self-supervised learning, and graph-based models applied to complex, weakly labeled data. My work spans the full lifecycle from model design and research to production deployment in banking systems under strict GDPR and reliability constraints.

Projects

Atlas

Large Financial Model
08/2023 – present
  • •Designed a self-supervised foundation model for transactional data, conceptually similar to LLMs but tailored for financial graphs
  • •Architecture based on temporal graph learning, enabling continuous representation updates as new transactions arrive
  • •Severe label scarcity, heterogeneous data sources, and regulatory constraints
  • •Implemented from scratch in PyTorch, including training pipelines and inference infrastructure
  • •Deployed in private cloud, serving real-time predictions and forming the analytical backbone of AI Explore

    • AI Explore

    AI-Driven AML Platform
    12/2022 – present
  • •Personally initiated and launched a new AI-based AML data and analytics platform
  • •Designed the platform’s core analytical concepts, centered around customer embeddings and unsupervised learning
  • •Built a visual “customer map” enabling investigators to explore behavior, risk patterns, and anomalies
  • •Integrated Atlas embeddings with downstream analytics and investigative workflows
    • Skills
    Machine Learning & AI
    • •Deep Learning (PyTorch)
    • •Self-Supervised Learning
    • •Graph & Temporal Models
    • •Representation Learning & Embeddings
    • •Time Series Modeling
    • •Unsupervised & Weakly Supervised Learning
    • •Agentic Automation & Development
    Mathematics & Foundations
    • •Algebra (Linear & Abstract)
    • •Probability & Statistics
    • •Graph Theory
    • •Optimization
    • •Topology & Topological Data Analysis
    Engineering & Systems
    • •Python, SQL
    • •MLOps
    • •Real-time & Batch Inference
    • •Docker, Git
    • •Databases (ClickHouse, Postgres)
    • •Cloud & Private Cloud Environments
    Professional Experience

    Data Scientist / ML Engineer

    TietoEvry
    12/2022 – present | Bergen, Norway
  • •Lead development of Atlas, a self-supervised foundation model for financial crime detection based on temporal graph learning
  • •Define ML strategy and core modeling approaches for next-generation AML analytics platforms
  • •Design models for weakly labeled, high-dimensional transactional data under strict GDPR constraints
  • •Develop and deploy real-time ML systems integrated into production banking environments
  • •Translate research and modeling results into operational systems used by European banks
  • •Communicate technical concepts to executives, customers, and cross-functional engineering teams
    • Sigurd Roll Solberg
    1 / 2
  • •Platform is currently piloted with multiple major Norwegian banks

    • Internal LLM Service

    Internal LLM Platform
    02/2025 – present
  • •Initiated project for secure OpenShift-hosted platform for internal LLM deployment and management
  • •Controlled experimentation with confidential data via access controls, on-prem hosting, and isolated execution
  • •Multi-team support with focus on reliability, security, and operational usability
    • Education

    University of Bergen

    Master´s Degree in Machine Learning
    2021 – 2023 | Bergen, Norway

    Focus on topological methods in machine learning. Master’s thesis on "Vectorizing Distributed Homology with Deep Set of Set Networks". Coursework in data science, machine learning, mathematics and algorithms.

    GPA: 3.71

    University of Bergen

    Bachelor’s Degree in Theoretical Mathematics
    2018 – 2021 | Bergen, Norway

    Focus on graph theory and clustering. Bachelor’s thesis on minimum spanning trees in hierarchical cluster analysis. Coursework in linear algebra, discrete mathematics, probability, statistics, and algorithms.

    References
    Nello Blaser, Associate Professor, University of Bergen
    [email protected], +47 55584791
    Pekka Parviainen, Associate Professor, University of Bergen
    Nikolay Martyushenko, Head of AI, TietoEvry Banking
    [email protected], +47 47379550
    Øystein Namtvedt, Head of Engineering, TietoEvry Banking
    Gurvinder Singh Dahiya, Lead Infrastructure Architect, TietoEvry Banking
    [email protected], +47 45069504
    Dan Henrik Sekse, Senior Researcher, Norwegian Defence Research Establishment

    Summer Intern (2 Summers)

    Norwegian Defence Research Establishment
    06/2020 – 08/2022 | Horten, Norway
  • •Simulation and data-processing tools for maritime and seabed analysis, including sensor and sonar visualisation
  • •Worked under formal security clearance procedures in a controlled research environment
  • •Technical reporting and presentation of design and modeling results

    • Student Teaching Assistant

    University of Bergen
    2020 – 2022 | Bergen, Norway
  • •Teaching assistant for Statistics, Algorithms & Data Structures, and Machine Learning
  • •Led group sessions, exam preparation, grading, and end-of-semester crash courses
    • Organisations

    YoungAI

    Board Member & Event Manager
    2023 – present | Bergen, Norway
  • •Board member of a volunteer AI/​ML community for young professionals
  • •Planning and execution of talks, workshops, and networking events
  • •Speaker on technical topics (e.g. equivariance in temporal graphs) and career-oriented sessions
  • •Ongoing responsibility for community activity and continuity

    • Awards

    CEDAS Hackaton Winner

    CEDAS & Eviny
    02/2025
  • •Designed and implemented a GRU-based model for predicting electric vehicle charging behavior using real-world utility data
  • •Combined time-series forecasting with embedding-based pattern analysis and clustering
  • •Recognized for technical depth and predictive performance
    • Languages
    Norwegian — Native/Bilingual
    English — Fluent
    French — Basic
    Sigurd Roll Solberg
    2 / 2