Dorina
Ababii

Research Analyst → Financial Analyst → Junior Portfolio Manager

HCP Asset Management · Zürich / Geneva, Switzerland

• [ INVESTMENTS ]
• [ DATA SCIENCE ]
• [ PORTFOLIO RISK MANAGEMENT ]
• [ PORTFOLIO MANAGEMENT ]

• Built a systematic equity fund from scratch: research, machine learning model selection, and live portfolio construction across three AMCs, together raising about CHF 15M in their early fundraising phase, delivering ~24% aggregated net gain (through my January 2026 departure).
• Adapted established machine learning methods to the fund's investment process rather than building new ones, working backward from what a portfolio manager actually needs day to day to make a decision. Used Bloomberg for data and Python for analysis and automation, delivering a fully automated pipeline from research idea to live portfolio decision.
• Built automated Python pipelines for data sourcing, feature engineering, model validation, and reporting, so the same analysis could be rerun and checked step by step instead of redone by hand each time.
• Presented investment results directly to institutional investors, translating quantitative models into decision frameworks non-specialists could act on. This helped onboard and retain clients by teaching them how the strategy actually worked.

Credit Analyst Intern

Mobiasbanca Groupe Société Générale · Chisinau, Republic of Moldova

• Collected and verified KYC (Know Your Customer) documentation for corporate and individual clients, checking financial and personal information against regulatory requirements.
• Helped migrate client records into a new data-automation tool the bank was rolling out, entering and validating data as part of that transition.
• Supported two client-facing conversations directly in the client's own language, requesting and reviewing the documents needed for KYC checks.
• Worked alongside senior credit analysts on credit risk assessments for corporate clients, analyzing financial statements, credit history, and market conditions to evaluate creditworthiness.

Accounting Intern

Vremealux · Chisinau, Republic of Moldova

• Managed the generation of monthly invoices and ensured timely processing of accounts payable and accounts receivable;
• Conducted financial data analysis and prepared balance sheets and income statements for senior management to aid in financial reporting and decision-making;
• Processed employee payroll using 1C Enterprise accounting software, ensuring accuracy and adherence to payroll schedules.

CAS in Risk Management

University of Zürich · Zürich, Switzerland · GPA: 6/6

Course Objective

• Focused on identifying, measuring, and mitigating financial and non-financial risks in banking and finance;
• Covered current challenges in risk governance, sustainability, and digital assets management.

MSc. Banking and Finance | Capital Markets & Data Science

Zürich University of Applied Sciences · Zürich, Switzerland

Course Objective

• Focused on advanced quantitative skills, including quantitative investment strategies, machine learning, and risk management;
• Leveraged data science techniques to optimize decision-making in capital markets.

BSc. Economics

University of Rouen Normandie · Rouen, France

Course Objective

• Gained in-depth knowledge of theoretical and empirical economic principles;
• Focused on microeconomics, macroeconomics, econometrics, and statistics;
• Developed an understanding of economic behavior, policy, and market dynamics.

BSc. Economics (1st Year)

University of Reims Champagne-Ardenne · Reims, France

Course Objective

• Studied core economic principles, focusing on microeconomics, macroeconomics, and financial markets;
• Explored the relationship between economic policies and their societal impacts;
• Laid the foundation for understanding strategic management in economics and business contexts.

High School, Profile in Sciences and Accounting

CNC of ASEM · Chisinau, Republic of Moldova

Proficiencies Acquired

• Studied science subjects (math, physics, chemistry) along with professional accounting principles;
• Gained hands-on experience in bookkeeping, financial statements, and auditing procedures.

SICTIC — Startup Jury Scoring Framework

SICTIC · Private Equity, Venture Capital

Building the scoring framework behind SICTIC's startup jury process: 231 due-diligence questions across four startup stages and eight industry sectors, still to be tested against 71 realistic use cases. Built using a multi-agent workflow: Hermes and MiniMax for initial research and drafting, Claude for orchestration and development, and Codex for testing, all connected through GitHub. Now extending it with the review team's jury-rating criteria.

Restaurant Operations Platform

Personal Project · Web Development, Operations

Turned a restaurant's ad-hoc, Excel-based profit tracking into a real web application: a waiter interface for tables and orders, plus a kitchen Telegram bot for stock tracking and demand forecasts. Built the requirements directly from the restaurant's own day-to-day floor and kitchen workflow, using the same multi-agent workflow as the SICTIC project (Hermes and MiniMax for research, Claude for orchestration and development, Codex for testing). User accounts are live, and the platform now covers 74 staff-facing actions across ordering, stock, and kitchen operations.

Personal Resume Website

Personal Project · Web Development, LLM Integration

Designed and built a personal resume website starting October 2024, updated periodically since rather than continuously, most recently through the same multi-agent workflow as my other projects (Hermes and MiniMax for research, Claude for orchestration and development, Codex for testing). Visitors can ask the chat assistant questions grounded in my real background, check job fit against a pasted job description, browse a chronological career timeline, download my CV or the conversation, and reach me through a contact form. Deployed on Vercel.

How can AI be leveraged to enable CRO employees to work more efficiently?

Winner, Team RiskON @ University of Zürich · CRO, Compliance

RiskON 2024 Hackathon Challenge 3, proposed by Julius Bär. Built a proof-of-concept using Python and DistilBERT (a language model) to automatically compare old and new regulatory documents in English and German and flag substantive changes, addressing real multi-jurisdictional compliance costs (Basel III, FINMA, MiFID II, GDPR). Wrote the resulting chapter, published in a University of Zürich Department of Finance report: Innovation in Risk Management: Three AI-Enabled Solutions for Swiss Banking ↗.

Master's Thesis: Machine Learning in Asset Pricing

Zürich University of Applied Sciences · Asset Management, Portfolio Management, Data Science

Explored the use of OLS, LSTM, and CatBoost for asset pricing, specifically in regional and global portfolio management. Executed experiments using international stock data from 2000-2021, split into development and testing phases. Gathered stock data from Refinitiv and macroeconomic indicators from FRED, applying Pearson-Correlation for feature selection. Demonstrated that deep learning models, like LSTM, outperform linear models in predictive accuracy and highlighted the role of explainable AI in improving portfolio management strategies. The clearest lesson: data quality matters more than data quantity, and locally, machine learning can help build alpha-generating portfolios, even though beating the market globally is much harder once you factor in geopolitics, currency, and monetary policy. Read the thesis ↗

Text Classification and Analysis

Zürich University of Applied Sciences · Sentiment Analysis, Natural Language Processing (NLP)

Developed and evaluated multinomial Naive Bayes and neural network models for text sentiment analysis using a Kaggle dataset. Assessed model performance, focusing on accuracy and effectiveness in classifying sentiments. Demonstrated proficiency in Python for text processing and model development, producing a positive/negative sentiment classification matrix for evaluation.

Credit Card Fraud Detection

Zürich University of Applied Sciences · Risk Management, Compliance, KYC

Co-authored a research paper (with Tsuyoshi Iwata, supervised by Dr. Bledar Fazlija) on making black-box machine learning models interpretable in finance, using a credit-card fraud dataset (590,540 e-commerce transactions from Vesta Corporation, via Kaggle) as the case study. Trained two interpretable models (logistic regression, a shallow decision tree) and one black-box model (a 1D convolutional neural network), and showed the black-box model outperformed both on prediction accuracy. Applied three interpretation methods (feature importance, partial dependence plots, LIME) to the neural network to make that advantage usable: feature importance and partial dependence plots for a credit-card provider to check the model's trustworthiness, partial dependence plots and LIME to explain to customers why a transaction was declined. Read the paper ↗ View the code ↗

Testing for Heteroskedasticity, Multicollinearity, Serial Autocorrelation

Zürich University of Applied Sciences · Statistics, Data Analysis

Collected and organized stock and economic data to build a comprehensive database for analysis. Conducted OLS regression to analyze stock returns, optimizing models based on R² values. Ensured accuracy and reliability by testing for heteroscedasticity, normality, autocorrelation, and addressing multicollinearity. Demonstrated strong skills in statistical analysis and data integrity in financial modeling. View the code ↗

Switzerland's Financial System and Real Economy Interplay During the Pandemic

Zürich University of Applied Sciences · Statistics, Data Analysis

Co-authored a group research paper analyzing how COVID-19 affected the interaction between Switzerland's real economy and its financial system: a real business cycle model on GDP and employment, plus Swiss government bond yields, credit risk indicators, inflation, and stock market impact. Concluded no financial crisis was imminent in the near term, provided Swiss banks' loan quality held up. Read the paper ↗

CFA Challenge

Zürich University of Applied Sciences · Statistics, Data Analysis · University level competition

Conducted a comprehensive financial analysis of a single stock to assess its buy, hold, or sell potential. Applied financial theories and valuation techniques to evaluate the stock's performance and risk profile. Demonstrated proficiency in financial modeling and risk assessment, providing clear investment recommendations in a competitive academic environment.