Research Assistant Malmö University

Alberto
Biscalchin

Applied AI, Data Science and agentic systems for the analysis of human behaviour.

I build agentic AI systems and reproducible computational pipelines to study travel behaviour and urban data. I combine applied machine learning and statistical evaluation with backend prototyping, decision-support systems and teaching experience.

Download the full CV Contact

Portrait of Alberto Biscalchin in a burgundy suit against a charcoal background
Alberto Biscalchin

Profile

Research Assistant at Malmö University, with an MSc in Computer Science specialising in Applied Data Science and a BSc in Computer Science and Automation Engineering. I design agentic AI systems, analyse data on human behaviour and develop reproducible computational pipelines.

I combine applied machine learning and statistical evaluation with backend prototyping, decision-support systems and teaching experience.

Research interests

Trustworthy and reproducible AI, human–AI collaboration, computational modelling of behaviour and visual analytics. I am interested in extending this work towards multimodal behavioural sensing, eye tracking and reinforcement learning, including how people adapt strategies, attention and trust when interacting with autonomous AI systems.

Research

Modular agentic systems, statistical analysis and reproducible pipelines for behavioural and urban data.

Research Assistant

Malmö University, Department of Technology and Society Sweden 2024 — present

  • Research on AI-assisted analysis of travel behaviour, mobility choices and urban planning data, including large-scale surveys and demographic differences.
  • Development of Python pipelines for data cleaning, feature selection, statistical analysis, model-supported reasoning, retrieval and report generation.
  • Design of modular agentic systems with specialised components for preprocessing, analysis, classification, validation and reporting.
  • Use of structured interaction traces, shared workflow states, validation checkpoints and intermediate outputs to support interpretability, auditing and reproducibility.
  • Contribution to study design, expert and participant protocols, rating instruments, statistical validation and transparent reporting of limitations.
  • Development of retrieval-augmented systems for urban planning and mobility decision support; co-supervision of a master's thesis on multimodal RAG for travel behaviour analysis.

Master's thesis

Leveraging LLMs for Understanding Travel Behaviour

A case study on older adults in Malmö

Malmö University 2025 grade A

  • Design and implementation of a multi-agent system based on foundation models for large-scale travel behaviour analysis, with structured JSON handovers between specialised agents.
  • Evaluation of the system in a blind study with six domain experts, compared with a commercial single-agent system and a baseline written by human experts.
  • Analysis of correctness, reasoning, relevance and clarity ratings using one-way ANOVA and Tukey HSD post-hoc tests.
  • Grade A, Thesis of the Year award and Wollmarska fonden scholarship 2026; the work was developed into a peer-reviewed paper.

Publications

Two peer-reviewed contributions and three manuscripts spanning applied AI, urban mobility and learning.

Peer-reviewed conference proceedings

  1. IEEE SoftCOM Split, Croatia 2025

    Designing Education for the AI Era: Principles for Integrating LLMs in Pedagogy

    Biscalchin, A., Sarkheyli-Hägele, A., Eriksson, J., Vogel, B.

    doi.org/10.23919/SoftCOM66362.2025.11197458

  2. FLLM Vienna, Austria 2025

    Multi-Agent Foundation Models for Urban Mobility: The Malmö Elderly Case

    Biscalchin, A., Sarkheyli, E., Sarkheyli-Hägele, A.

    doi.org/10.1109/FLLM67465.2025.11391006

Preprints and manuscripts

  1. Preprint 2026

    Large Language Models in Travel Behaviour Research: A Systematic Mapping of Applications, Data, and Emerging Practices

    Biscalchin, A., Sarkheyli-Hägele, A., Sarkheyli, E., Lord, J., Gołdyka, K.

  2. In preparation — British Journal of Educational Technology 2026

    The Temporal Paradox of AI-Assisted Learning

    Biscalchin, A., Sarkheyli-Hägele, A., Eriksson, J., Vogel, B., Björklund Boistrup, L.

  3. In preparation — Information Systems 2026

    A Reproducible Pipeline to Discover Serious Games

    Biscalchin, A., Spalazzese, R., et al.

Teaching and professional experience

  1. 2024 — present

    Research Assistant

    Malmö University, Department of Technology and Society — Sweden

    Research on AI and mobility, data pipelines and agentic systems; co-supervision of a master's thesis.

  2. 2024 — 2025

    Teaching Assistant

    Malmö University — Sweden

    Support for undergraduate teaching and student learning activities.

  3. 2022 — 2023

    Secondary school teacher

    Istituto Juvarra — Italy

    Mathematics, Physics and Computer Science; more than 500 teaching hours.

  4. 2021 — 2023

    Head Teacher

    Centro Studi Omnibus — Italy

    Mathematics, Physics and Computer Science.

  5. 2018 — 2019

    Full-stack Developer

    GIGA srls — Italy

  6. 2014 — 2015

    Head of IT Department

    Mondadori Mega Store — Italy

  7. 2012 — 2018

    Freelance photographer and videomaker

    Italy

  8. 2011 — 2012

    Customs consultant

    FCA — Italy

  9. 2011

    Industrial quality control

    Con.Top S.R.L. — Italy

Education

  1. 2025

    MSc in Computer Science — Applied Data Science

    Malmö University — Sweden 120 credits

    Master's thesis with grade A. Relevant courses: Advanced Machine Learning (A); Big Data Analytics on Cloud Computing Infrastructures (B); Statistical Methods for Data Science (B); Research Methods of Computer Science and Fundamental Computational Theory (B).

  2. 2025 — 2026

    Further second-cycle courses in Computer Science

    Malmö University — Sweden 30 ECTS

    IoT Prototyping and User Experience (15 ECTS, A); Introduction to IoT (15 ECTS, B).

  3. 2022

    BSc in Computer Science and Automation Engineering

    Università eCampus — Italy 180 ECTS

    Bachelor's degree, Department of Computing and Control Systems.

  4. 2019

    Executive Master in Digital Video Editing

    MyWeb School, Naples — Italy

  5. 2010

    High school diploma in Applied Sciences

    Istituto Giulio Natta, Rivoli — Italy

Technical and methodological skills

Programming and ML
  • Python
  • PyTorch
  • Transformers
  • Pandas
  • C/C++
  • MATLAB

Classification, feature engineering, benchmarking, model evaluation and error analysis.

Agentic AI
  • Multi-agent architectures
  • Foundation models
  • Prompt-controlled agents
  • RAG
  • Structured tool use
  • Intermediate state validation
  • OpenAI API
  • LangChain
  • CrewAI
Data and backend
  • SQL
  • PostgreSQL
  • FastAPI
  • Qdrant
  • CSV/JSON processing
  • Schema-aware pipelines
  • Docker
  • Git
  • GitHub
  • CI/CD
Statistics and evaluation
  • Descriptive statistics
  • One-way ANOVA
  • Tukey HSD
  • Inter-rater agreement
  • Controlled comparisons
  • Expert evaluation
  • Reproducible reporting
Human-centred methods and visualisation
  • Participant studies
  • Likert-scale instruments
  • Human-in-the-loop
  • Research ethics
  • QGIS
  • Matplotlib
  • Spatial analysis
  • Multimodal retrieval
Tools
  • AutoCAD
  • Inventor
  • Adobe Suite
  • Reaper
Languages
  • Italian — native
  • English — C1, IELTS Academic 7.0
  • Spanish — B2

Selected projects

  • Multimodal RAG for travel behaviour analysis

    Contribution to a hybrid retrieval system combining layout-aware text processing with region-based visual retrieval. Evaluation used a benchmark of 487 questions covering correctness, faithfulness, latency, token efficiency and inter-rater agreement.

  • MAU Advisor

    Retrieval-augmented platform for urban planning support: structured knowledge retrieval, backend services, domain modelling and user-oriented reporting.

  • Pilot study on AI-assisted learning

    Design and conduct of a study comparing traditional reading with AI-assisted study via NotebookLM, with timed tasks, open questions and analysis of answers and learning strategies.

  • Evidence synthesis

    PRISMA-guided screening and data extraction for a systematic mapping of LLMs in travel behaviour research. Contribution to a longitudinal review on AI-assisted learning: regulation, strategy development, dependence and long-term learning trajectories.

Awards and academic service

  • 2026

    Thesis of the Year, Malmö University

  • 2026

    Scholarship Wollmarska fonden

  • —

    Reviewer for Mobile Networks and Applications

Dissemination

Conference presentations and research dissemination activities in applied AI, mobility analysis and AI-assisted learning.

References

Academic references available on request.

Contact

Email

alberto.biscalchin@gmail.com

Institutional: alberto.biscalchin@mau.se

Malmö, Sweden

Download the original CV

Full CV in PDF, with publications, teaching and references.