Quantitative Finance · Data · Software
01 · About
I’m Raphaël Catenacci, an engineering student at ENSAE Paris, part of the Institut Polytechnique de Paris, graduating in 2027. I am currently completing the final year of a Master’s degree in Financial Engineering at the Université de Montréal in Canada.
I am passionate about quantitative finance, machine learning, and software development. I build data-driven financial tools, from automated market-data pipelines and portfolio-risk models to pricing engines and machine-learning research workflows.
Before moving into finance and technology, I competed as a high-level BMX Racing athlete, reaching 9th in the U18 UCI World Ranking in 2019. This experience shaped the discipline, resilience, and performance mindset that I bring to every technical and quantitative project.
Download my CV →02 · Background
Exchange / Academic Program · Montréal, Canada
Academic experience in a leading North American research environment, complementing my quantitative training through applied data science, optimization and financial modelling.
Relevant coursework: Stochastic Programming and Dynamic Pricing.
Engineering School · Palaiseau, France
Engineering degree at the Polytechnic Institute of Paris, with a focus on quantitative methods, statistics, financial mathematics and machine learning.
Relevant coursework: Financial Mathematics, Stochastic Processes, Machine Learning, Financial Instruments, Time Series Analysis, Statistical Simulation and Monte Carlo Methods.
03 · Experience
Quantitative Researcher · Part-time · Monaco / Remote
Developing an end-to-end algorithmic trading platform, from financial-data ingestion and storage to strategy research, backtesting and performance monitoring.
Designing a Bronze, Silver and Gold data lakehouse to standardize market, macroeconomic and alternative data while maintaining metadata, data quality and lineage. Integrating specialized AI agents to extract and structure information from unstructured sources, including financial reports, audio and video.
Building reusable quantitative research pipelines for feature engineering, signal generation, portfolio construction, risk analysis and strategy evaluation.
Python · Parquet · Apache Iceberg · DuckDB · Polars · Dagster · SeaweedFS/S3 · Docker
Quant Researcher · Off-cycle · Monaco
Developed a fully automated, provider-agnostic financial-data infrastructure for scalable market-data ingestion and processing, integrated with a web-based quantitative application.
Automated financial-instrument lifecycle management, including corporate actions, pricing updates, event processing and portfolio synchronization across trading workflows.
Implemented risk calculations, stress testing, exposure measurement and pricing models for trading portfolios, with cross-asset and cross-strategy support.
Developed local-LLM workflows for macroeconomic and alternative data cleaning, normalization, field mapping and pattern detection. Implemented sampling, meta-labelling and sample-weighting methodologies to improve predictive signal quality before machine-learning deployment.
Python · Bloomberg · Django · PostgreSQL · FastAPI · Uvicorn · QuantLib · JavaScript · Financial ML · Risk Modelling
Data Scientist Intern · Paris, France
Developed machine-learning algorithms within the Financial Compliance team to support financial-data analysis and compliance workflows.
Built an artificial neural-network solution using the Transformers library and applied data-processing and machine-learning techniques to financial compliance problems.
Python · PyTorch · Transformers · Pandas · Scikit-learn
04 · Expertise
Python, SQL, JavaScript, HTML, CSS, R, Git.
Quantitative research, market data, portfolio analytics, risk, pricing and financial modelling.
Data pipelines, APIs, databases, web applications, automation and machine learning.
Django, Numpy, Pandas, Polars, Scikit-learn, PyTorch, Transformers, CatBoost
05 · Selected work
Quantitative Finance
Designed and implemented multiple ML-driven trading strategies, evaluated with Sharpe ratio, drawdowns, and other risk-adjusted metrics.
Machine Learning
Built a CatBoost model to forecast French national gas consumption, leveraging AutoML and hyperparameter tuning for robust time-series predictions.
Finance App
Revolut portfolio scanner providing risk and performance analysis, personalized market news, and an economic calendar.
Open app ↗06 · My company
RC is a solo-owned technology studio specializing in bespoke software
solutions that make complex workflows simple, reliable, and measurable.
Our mission is to help teams regain time and clarity by designing intelligent data pipelines, automating high-friction processes, and building fully customizable interfaces tailored to real operational needs.
From quantitative research and financial operations to internal tooling and process automation. Our approach combines deep technical expertise with a strong understanding of business and compliance constraints.
End-to-end ingestion, cleaning, enrichment, and orchestration of structured and unstructured data, with robust monitoring and lineage tracking.
Scripted and event-driven automation of repetitive or error-prone processes across desktop, web, and backend systems to reduce manual effort and operational risk.
Integration of local or cloud-based AI models to enable conversational interfaces, document understanding, and multi-agent coordination within your custom software stack.
07 · Services
Financial, economic and alternative data pipelines designed for reliability, scalability and reuse.
Research tools, portfolio analytics, risk calculations, pricing models and backtesting environments.
Tailored web applications, APIs, automation workflows and business tools adapted to your processes.
Reach out to discuss your project and make a tailored reality.
08 · Case study
Quantitative data infrastructure & lakehouse architecture
During my off-cycle experience at Jukoï Capital in Monaco, I worked on the design and development of a provider-agnostic financial data infrastructure supporting quantitative research, market-data processing, portfolio workflows, risk calculations and pricing models.
The initial project focused on automating the ingestion, transformation and normalization of market, macroeconomic and alternative data. The resulting components were integrated into a web-based financial application, connecting research workflows with operational portfolio and risk-management tools.
I am now contributing to the design of a complete quantitative lakehouse architecture intended to make the platform more reproducible, observable and production-ready. The architecture separates raw, enriched, staging, canonical and gold data layers, while preserving source objects, provider vintages, immutable releases and replayable transformations.
The target platform combines Ceph RGW for sovereign S3-compatible object storage, Apache Iceberg for transactional analytical tables and Lakekeeper as the central Iceberg REST catalog. Trino, Spark, Polars and DuckDB provide shared SQL, distributed processing and local iteration, while Dagster orchestrates partitioned data assets and enforces data contracts.
A major focus of the architecture is financial reproducibility. Critical datasets distinguish between event time, publication time and verified availability time. Point-in-time feature construction and anti-look-ahead tests ensure that backtests only use information that would genuinely have been available at the time of each historical decision.
The platform also introduces controlled AI-assisted workflows. Local language models can support document extraction, data cleaning, normalization, field mapping and pattern detection, but their outputs remain restricted to enriched or staging layers until they pass deterministic validation, quality checks and, where required, human review.
The broader system includes lineage tracking, data-quality contracts, observability, agent evaluation, vector-search synchronization and disaster-recovery procedures. The objective is not to reproduce a monolithic data platform, but to build a modular and auditable foundation adapted to quantitative research and portfolio operations.
Build a scalable and provider-agnostic financial data platform capable of handling market data, macroeconomic data, alternative data, portfolio processes and quantitative research while preserving reproducibility, security and temporal correctness.
Designed a layered quantitative lakehouse based on Ceph RGW, Apache Iceberg and Lakekeeper, with Dagster orchestration, Trino and Spark processing, declarative data contracts, OpenLineage tracking, Feast point-in-time features and controlled MLflow-based AI evaluation.
Established a reusable foundation for certified datasets, reproducible backtests, auditable research releases and more reliable connections between data engineering, quantitative research, trading and portfolio-management workflows.