Projects

Work projects and research

Work projects

Social signal intelligence for interactive AI systems

Company

Interhuman AI

Summary

Behavioral AI work focused on building models that detect and respond to human interaction dynamics.

  • Designed annotation framework to label conversational social signals (agreement, hesitation, confusion, engagement).
  • Built ground-truth aggregation pipeline combining expert and crowd annotations for model training.
  • Developed feedback layer mapping detected signals into adaptive coaching responses and avatar behavior.
Social IntelligenceHuman-AI InteractionConversational Analytics

Methods: Multi-annotator labeling, signal taxonomy design, behavioral feedback mapping

Modeling self-conscious decision-making in London contract environments

Company

Inhalation Dev

Summary

Behavioral modeling project on how self-consciousness shapes high-stakes contract decisions in the London commercial property market.

  • Developed a conceptual and computational framework to simulate how emotional tension, identity investment, and cognitive load influence contract-signing behavior.
  • Integrated research from social psychology and consumer behavior on self-consciousness, impression management, and socially sensitive purchasing decisions.
  • Designed a modular agent-based model structure allowing adaptation across industries (commercial property, finance, tech) and contract types (traditional lease, serviced office, managed office).
Behavioral ModelingAgent-Based ModelingDecision Science

Methods: Literature synthesis, conceptual modeling, agent design framework, cross-domain generalization

AI-driven risk intelligence from large-scale news and signal data for strategic decisions

Company

Emerging Trajectories

Summary

Applied AI and market intelligence research to anticipate risks, opportunities, and macro shifts for clients including HP, GoodMills, and Morgan Stanley.

  • Designed structured keyword-discovery workflow to retrieve high-signal news from large media APIs and reduce topic noise.
  • Built prompt frameworks to turn raw article streams into scenario hypotheses and forward-looking narratives.
  • Produced sector-specific outlooks combining geopolitical events, economic indicators, and behavioral drivers for client briefings.
Behavioral ForesightMarket IntelligenceAI Analysis

Methods: News data retrieval, prompt design, scenario synthesis, signal filtering

Detecting industry opportunities and best-fit growth areas for GTM strategy

Company

Toptal

Summary

Mandate to identify untapped industries and high-value prospects for go-to-market execution.

  • Built a data-driven industry evaluation model using 22 metrics from sources including US BLS, Lightcast, and BlackRock to score industry attractiveness and ability to win.
  • Produced industry scorecards and prospect-needs discovery outputs covering outlook, strategic priorities, technological needs, and SWOT.
  • Prioritized Healthcare and Life Sciences, Financial Services, and Industrials for prototyping; work informed outreach campaigns and launch of a new consulting offer.
GTM StrategyMarket IntelligenceIndustry Analysis

Methods: Industry scoring model, account prioritization, prospect-needs discovery

Improving micro-mobility behavior

Summary

Behavioral intervention design with a micromobility company to reduce parking violations.

  • Ran a survey of 500 riders in Germany to identify behavioral drivers of non-compliant parking and define target segments.
  • Designed two app interventions based on social norms and care priming (context-specific parking pop-ups).
  • Achieved a 12% reduction in measurable parking violations and reduced local authority claims.
Behavioral ScienceUX ResearchChoice Architecture

Methods: Survey research, segmentation, intervention design

Increasing credit card take-up rate

Summary

Behavioral experimentation with a banking partner to improve credit-card adoption.

  • Executed two randomized online studies (n=1,000 each, EU 18+) via Qualtrics to test product and information design changes.
  • Study 1 tested card color (red vs black); Study 2 tested information framing with and without a non-informative positive image.
  • Preliminary outcomes showed +11.8% hypothetical take-up for black cards and +4.8% for image-supported communication.
Consumer BehaviorExperiment DesignFinancial Services

Methods: Randomized survey experiments, linear and logistic regression (Matlab, Stata)

Research & writing

Behavioral macroeconomics, climate and development economics, agent-based modeling, household finance, and misinformation.

The leverage self-delusion: perceived wealth and cognitive sophistication

Authors

Assenza, Tiziana; Cardaci, Alberto; Delli Gatti, Domenico

Publication

Journal of Economic Interaction and Coordination

2025

Links

Laboratory evidence shows participants misjudge net worth based on leverage composition even when objective wealth is identical. Misperception is associated with lower cognitive sophistication and stronger consumption responses to income shocks, highlighting perception bias in financial behavior.

Laboratory ExperimentBehavioral MeasurementCognitive Reflection TestWealth PerceptionBehavioral Finance

Fake News: Susceptibility, Awareness and Solutions

Authors

Assenza, Tiziana; Cardaci, Alberto; Huber, Stefanie

Publication

Toulouse School of Economics Working Paper

No. 1519 (2024)

Links

RCT with a fake-news detection task and personalized feedback shows participants overestimate their ability to identify misinformation. Learning true susceptibility increases willingness to pay for protection, supporting demand-side policy interventions based on belief updating.

Randomized Controlled TrialBelief UpdatingInformation-Through-ExperienceMisinformation EconomicsPolicy Design

Consumption and Account Balances in Crises: Have We Neglected Cognitive Load?

Authors

Assenza, Tiziana; Cardaci, Alberto; Haliassos, Michael

Publication

Toulouse School of Economics Working Paper

No. 1499 (2023)

Links

Randomized online experiment on French households studies consumption and savings decisions under income risk. Cognitive load reduces consumption and increases balances because decision rules become less optimal, revealing limits of standard intertemporal consumption models under high mental load.

Randomized Controlled TrialBehavioral ExperimentIncentivized Online ExperimentCognitive LoadHousehold Finance

Inequality and imbalances: a monetary union agent-based model

Authors

Cardaci, Alberto; Saraceno, Francesco

Publication

Journal of Evolutionary Economics

29, 853-890 (2019)

Links

A two-country ABM explains Eurozone-type imbalances: inequality plus credit access drives debt-led growth in borrowing countries and export-led growth in lending countries. Cross-border lending sustains demand until sudden stops trigger crisis and balance-of-payments divergence.

Multi-Country ABMMonetary Union SimulationCapital FlowsCurrent Account ImbalancesEurozone Crisis

Between Scylla and Charybdis: income distribution, consumer credit, and business cycles

Authors

Cardaci, Alberto; Saraceno, Francesco

Publication

Economic Inquiry

57(2), 953-971 (2019)

Links

A computational macro model with heterogeneous households shows two inequality-driven regimes: credit expansion produces debt-led booms and fragility, while constrained credit produces demand collapse and recession. Progressive taxation stabilizes the economy more effectively than countercyclical fiscal policy.

Computational MacroHeterogeneous AgentsCredit Market SimulationBusiness CyclesFiscal Policy

Inequality, household debt and financial instability: An agent-based perspective

Authors

Cardaci, Alberto

Publication

Journal of Economic Behavior and Organization

149, 434-458 (2018)

Links

Builds a credit-network ABM with heterogeneous households, peer-driven consumption, mortgages, and housing markets to reinterpret the 2007-08 crisis. Rising inequality triggers imitation-led consumption and debt accumulation, increasing non-performing loans and endogenously generating financial instability.

Agent-Based ModelingNetwork AnalysisHeterogeneous AgentsInequalityHousehold Debt

Policy experiments in an agent-based model with credit networks

Authors

Assenza, Tiziana; Cardaci, Alberto; Delli Gatti, Domenico; Grazzini, Jakob

Publication

Economics: The Open-Access, Open-Assessment E-Journal

12(2018-47)

Links

Extends a macro ABM with an interbank network and public sector to compare fixed rates, Taylor rule, helicopter money, and regulatory constraints. Taylor-rule policy reduces crisis frequency and duration, and leverage regulation outperforms liquidity regulation for system stability.

Agent-Based MacroeconomicsNetwork ModelingInterbank SimulationMonetary PolicyMacroprudential Regulation

Impact of rainfall pattern on cereal market and food security in Sudan: stochastic approach and CGE model

Authors

Sassi, Maria; Cardaci, Alberto

Publication

Food Policy

43, 321-331 (2013)

Links

Monte Carlo rainfall scenarios are integrated into a CGE model of Sudan to estimate food-security impacts. Lower rainfall reduces cereal output, raises prices, lowers income, and disproportionately harms poorer households, showing that food policy should integrate environmental risk with macroeconomic dynamics.

CGEMonte CarloSocial Accounting MatrixClimate EconomicsFood Security