Available for opportunities
AI/ML Engineer · Data Scientist · Builder
I design and build intelligent systems end-to-end — from research and experimentation to production-grade deployment. Based in Italy, working globally.
About
I'm Alessandro — an AI/ML Engineer and Data Scientist with three years of professional experience building end-to-end intelligent systems. My work sits at the intersection of rigorous research and pragmatic engineering.
My background spans the full ML lifecycle: from problem framing and data curation, to training, evaluation, and deploying models that hold up under real conditions. Recently I've deepened my backend engineering skills to better own the complete stack — because a model that can't be served is just an experiment.
I'm also a polyglot. I speak Italian, English, Spanish, Russian and Ukrainian, with German on the way. Language learning is a proxy for how I approach everything: systematically, with genuine curiosity.
Skills
End-to-end pipelines from raw data to production serving. Training, evaluation, model monitoring, drift detection.
Statistical analysis, hypothesis testing, experiment design. Turning ambiguous data into clear, defensible decisions.
Async APIs and the infrastructure that powers them. Built for maintainability, tested from day one.
Projects
End-to-end bioinformatics and ML pipeline to infer cancer stage from tissue DNA methylation profiles. Involved deep field research into cancer biology and staging, expert consultation, and processing 200GB+ of raw methylation data via Dask and Polars Lazy DataFrames — well beyond single-machine memory limits. Compared multiple model architectures (XGBoost, Scikit-learn, PyTorch) with full evaluation. Achieved 98% accuracy distinguishing cancer vs. cancer-free samples.
Comprehensive clinical data collection platform built for oncology specialists to digitise and streamline patient anamnestic intake. Designed to be minimalist and efficient, it integrates PedigreeJS for interactive graphical pedigree tree building and the CanRisk API for cancer mutation probability assessment. Role-based access is managed via Microsoft Entra with Azure deployment behind Microsoft IIS.
A collection of LLM-powered developer tools. Tailor-your-CV lets users align their CV to a specific job description — surfacing the most relevant experiences and highlights for recruiters — powered by OpenAI, Gemini, and local Ollama models, deployed as a Streamlit app. Local RAG indexes local files (PDF, Markdown, HTML, TXT and more) into a ChromaDB vector store with smart chunking, enabling precise semantic retrieval with full source attribution.
How I Work
Deep-dive into the problem before writing any code. I'd rather ask the uncomfortable question in week one than refactor the wrong architecture in week six.
Rapid prototyping with a bias toward observable, testable components. Every experiment tracked, every assumption documented.
Offline evals, A/B baselines, adversarial test sets. A model that can't be measured can't be improved and shouldn't be trusted.
Containerised, CI-integrated, observability from day one. Deployment isn't the end — it's when the real data collection starts.
Contact
Open to freelance projects, consulting engagements, full-time roles, and the occasional interesting side collaboration. Response within 24 hours.