Albert Queraltó, PhD · Barcelona, Spain

I build reliable machine-learning systems from real-world data.

Data scientist and software developer working across time-series modelling, anomaly detection, data pipelines, APIs, and full-stack decision tools. Currently applying these skills to environmental monitoring systems.

View selected work
Research
PhD-trained problem solving
Engineering
Maintainable APIs and full-stack tools
Applied ML
Real-time monitoring and anomaly detection
Production
FastAPI, React, PostgreSQL, Docker
Data systems
API, FTP, Airflow and geospatial pipelines

Core stack

  • Python
  • Pandas
  • NumPy
  • scikit-learn
  • PyTorch
  • TensorFlow
  • FastAPI
  • PostgreSQL
  • Docker
  • Linux
  • Bash
  • Git
  • React
  • TypeScript
  • Plotly
  • Power BI

How I work

From ambiguous problem to maintainable system.

I combine scientific reasoning with software engineering: clarify the decision, validate the data, build the smallest useful model, and make the result observable and usable in production.

Data and analytics Turn messy operational data into trustworthy signals.
  • Time-series cleaning, feature engineering, and exploratory analysis
  • Statistical modelling, validation, and interpretable reporting
  • Visualisation with Plotly, Bokeh, Power BI, and Streamlit
Machine-learning systems Develop models that survive contact with production data.
  • Supervised learning and deep learning with scikit-learn, PyTorch, and TensorFlow
  • Time-series prediction, anomaly detection, and interpretation with SHAP
  • Model-serving APIs, monitoring, reproducible training, and containers
Data platforms and pipelines Move data reliably from source systems to applications.
  • API, FTP, file, and scheduled ingestion workflows
  • PostgreSQL, PostGIS, Databricks, and analytical storage
  • Apache Airflow scheduling, ETL automation, and data-quality checks
Software engineering Package analytical work into maintainable products.
  • Python services with FastAPI and Flask
  • React and TypeScript interfaces for configuration and results
  • Docker, Linux, Git, testing, and clean architectural boundaries

Flagship projects

Production-minded SaaS for high-stakes business decisions.

Two larger systems that show the full path from data modelling and domain rules to APIs, background jobs, deployment, and decision-focused product design.

Flagship project · Machine learning SaaS

Payrithm turns invoice history into collection decisions.

An accounts-receivable intelligence platform that predicts late payments, estimates payment timing, forecasts incoming cash, and converts model outputs into prioritized collection workflows.

  • Uses chronological model validation to prevent future information from leaking into training.
  • Evaluates probability quality with ROC-AUC, calibration, and the Brier score.
  • Combines risk, invoice value, urgency, and reminder activity into an explainable collection-priority score.
  • Python
  • FastAPI
  • scikit-learn
  • React
  • PostgreSQL
  • Celery
  • Redis
  • Docker
View case study Read article Open application Source code
Payrithm accounts-receivable dashboard showing invoice risk and collection priorities

End-to-end ML product

Data ingestion, asynchronous processing, model training, prediction APIs, monitoring, and a decision-focused frontend.

Flagship project · Procurement SaaS

TenderWise turns public-procurement noise into bid/no-bid action.

A multilingual SaaS for European public procurement that ingests TED notices, scores opportunities against company profiles, separates strategic relevance from qualification readiness, and coordinates bid preparation.

  • Stores canonical TED notice history so source changes and failed-ingestion retries remain traceable.
  • Keeps relevance and qualification as separate decision axes before returning BID, REVIEW, or NO_BID.
  • Adds a team bid workspace with stages, internal deadlines, assignments, seeded checklists, comments, and notifications.
  • Runs as a Dockerized SaaS with workspaces, roles, billing limits, audit logs, backups, and optional queued AI briefs.
  • React
  • TypeScript
  • FastAPI
  • PostgreSQL
  • pgvector
  • Celery
  • Redis
  • Docker
  • Stripe
  • Ollama
View case study Read article Open application Source code
TenderWise procurement dashboard showing TED opportunity scores, qualification evidence, and bid workspace progress

Production procurement SaaS

React/Vite frontend, FastAPI API, PostgreSQL with pgvector, Redis/Celery workers, scheduled TED ingestion, billing, audit, backups, and optional AI briefs.

Experience

Applied data science in operational environments.

Experience building models, data pipelines, APIs, and technical tools for environmental monitoring, industrial systems, customer analytics, and scientific research.

  1. June 2025 — Present

    Software Developer

    Adasa Systems · El Prat de Llobregat, Spain

    • Implement backend functionality using Python and an IoT platform.
    • Integrate API, FTP, hydrological, meteorological, raster, and ensemble geographical data.
    • Build scheduled ingestion workflows with custom schedulers and Apache Airflow.
    • Store and expose operational data through PostgreSQL, PostGIS, GeoServer, and APIs.

  2. January 2023 — Present

    Data Scientist

    Adasa Systems · El Prat de Llobregat, Spain

    • Developed machine-learning models for real-time water-treatment and dam time-series data.
    • Built a contamination-event detector with more than 80% detection capability.
    • Used SHAP to interpret model predictions.
    • Created Streamlit and FastAPI tools for configuring, migrating, and visualizing models.
    • Developed a RAG application with ChromaDB, LangChain, Ollama, FastAPI, and React.

  3. June 2022 — January 2023

    Data Scientist

    HP through Between Technology · Sant Cugat del Vallès, Spain

    • Automated database ETL processes with Python and Databricks.
    • Analyzed repair, printing, customer-service, and review data.
    • Automated recurring reporting workflows with Python and PostgreSQL.

  4. January 2019 — June 2022

    Data Scientist and Researcher

    Institute of Materials Science of Barcelona · Cerdanyola del Vallès, Spain

    • Led high-throughput experimentation and data-driven material-discovery work.
    • Automated experimental hardware and data acquisition with Python.
    • Developed machine-learning models for experimental optimization and prediction.

Selected work

Projects that connect models, data, and software.

Machine-learning systems, analytical products, data pipelines, and web applications with detailed explanations of the decisions behind them.

4 of 9 projects

Payrithm application preview

ML SaaS · Financial operations

Payrithm

Deployed

An accounts-receivable intelligence platform that predicts late payments, estimates payment timing, forecasts cash receipts, and prioritizes collection work.

  • Python
  • FastAPI
  • scikit-learn
  • React
  • TypeScript
View case study Live preview
TenderWise application preview

Procurement SaaS · Decision support

TenderWise

Deployed

A multilingual procurement SaaS that ingests TED notices, scores opportunities against company profiles, separates relevance from qualification, and coordinates bid preparation.

  • React
  • TypeScript
  • Vite
  • FastAPI
  • PostgreSQL
View case study Live preview
Catalunya Weather Portal application preview

Weather planning portal

Catalunya Weather Portal

Deployed

An authenticated React portal for Catalonia weather planning, combining Meteocat station history, map-based alerts, air-quality signals, and activity recommendations.

  • TypeScript
  • React
  • Python
  • FastAPI
  • Leaflet
View case study Live preview

Start a conversation

Let’s build something useful.

Reach out about data science, machine-learning engineering, software engineering roles, or product collaborations that require a practical path from data and prototypes to production systems.

Based in Barcelona
Location
Barcelona, Spain