Antonio Ercole De Luca
AI/ML Software Engineer & Architect
With over 10 years as a Software Engineer, Machine Learning Engineer, Data Architect, and Scrum Master, I specialize in end-to-end AI and data solutions — from web scraping and database design to cloud orchestration, generative AI, and scalable ML pipelines.
The Journey
From NLP research in Pisa to building AI platforms across three continents — here are the chapters that shaped my career.
Where It All Started
I earned a BSc in Computer Science and MSc in Data Science & Business Informatics at the University of Pisa, researching IR, NLP, and Machine Learning. An Erasmus semester in Valencia broadened my perspective.
My first industry roles — at Synthema on EU-funded multilingual NLP pipelines, then at SpazioDati building Active Learning workflows — planted the seeds for everything that followed.
Going Nomad
Starting with an open-source LinkedIn scraper, I transitioned to full-time remote work via Upwork — delivering 5/5-star projects for 10+ clients while traveling through Europe, Peru, and the US.
Along the way I collaborated with KiwiBot at Berkeley SkyDeck, reconnecting with a contact from a nomadic camp in Lanzarote.
Robots & Burritos
At Berkeley SkyDeck, I helped KiwiBot integrate supplier APIs for their food delivery robot platform — handling 2,000+ orders in under two months. The startup spirit was contagious.
The Sabbatical
After years of consulting, I hit pause. I pursued a BSc in Statistics at the University of Palermo to strengthen my mathematical foundations — studying financial mathematics, multivariate analysis, and Bayesian inference.
During the 2020 lockdown, I contributed to Dask, implementing random sampling on distributed Bags. I wrote about it on Medium.
Building Community
In 2022, I relocated to Las Palmas and co-founded PyData Las Palmas, fostering innovation among digital nomads and expats. The island's vibrant tech community has been a perfect home base.
SunnyPlans — My Startup
SunnyPlans merges my passion for sustainability with solar industry innovation. I built an AI-powered geo-analytics platform helping businesses identify optimal terrains for utility-scale solar projects, using PostGIS and advanced algorithms.
My involvement is personal — it's a commitment to making solar energy mainstream.
About
I specialize in end-to-end AI and data solutions. My expertise spans the full data lifecycle — including web scraping, relational and NoSQL databases, cloud orchestration, RESTful APIs, and Jupyter Notebooks — while leveraging generative AI (LLMs, LangChain for agentic workflows) to boost efficiency.
I've built scalable AI-driven platforms in commodities trading, renewables, and finance. In 2019 I took a sabbatical to deepen my knowledge in Statistics and Finance, contributing to two scientific papers on Web Scraping and Semantic Crawling.
Education
- BSc Statistics & Data Science — University of Palermo
- MSc Data Science & Business Informatics — University of Pisa
- BSc Computer Science — University of Pisa
Languages
- Italian — Native
- English — Fluent
- Spanish — Fluent
Skills
Technologies and tools I work with daily.
Programming Languages
Cloud & DevOps / MLOps
AI / ML & Data Science
Frameworks & Libraries
Data Tools & Pipelines
Professional Skills
Experience
11 positions across AI, renewables, fintech, humanitarian aid, and more.
Architected scalable AI/ML pipelines and backend for fuel arbitrage platform, integrating generative AI for decision-making and Prophet/Bayesian models for gasoline tank demand forecasting.
- Developed load optimization algorithm reducing human error and fleet costs — 100% adoption by trading teams
- Built dbt/Dagster pipeline ingesting data from 329 stations and 1,000+ tanks
- Implemented TDD and CI/CD pipelines, reducing deployment errors by 80%
- Delivered Superset dashboard for mid-tier client in two weeks, ensuring retention
- Scraped 8 governmental websites for RFP data to feed the pipeline
Founded AI-powered geo-analytics platform for solar land selection using advanced algorithms and PostGIS for spatial data handling.
- Generated €3.1k revenue over 6 months from 2 clients
- Launched Chrome Extension for Stable Diffusion prompt engineering (200+ installs in first month)
- Built programmatic SEO project with sentiment analysis for restaurant rankings (100 daily visitors)
- Managed ETF portfolio with Bayesian forecasting, achieving 40% ROI over two years
Optimized real-time, parallel Windows software system for AI-driven food production.
- Processed ultrasound videos with FFmpeg; refactored features for efficiency
- Dockerized computer vision models with GPU access using CUDA and NVIDIA Docker
- Resolved concurrency bugs where FFmpeg competed with CV models for GPU resources
Launched digital mortgage platform; refactored codebase, set up environments, and implemented backward-compatible REST API changes with bug fixes.
Led 10+ engineer team in Scrum adoption; standardized TDD practices, identified technical debt, and established E2E automated testing foundations.
Launched and remotely managed a B&B-style Airbnb listing. Handled full remote customer care, pricing, guest communication, and problem solving across time zones.
- 100% response rate and 0% cancellation rate across 239 guest reviews
- 4.43 overall rating — communication 4.74, check-in 4.74, accuracy 4.7
- Managed operations while traveling Peru, US, and Southern Italy
Delivered 5/5-star projects for 10+ clients.
- Built zero-latency image search engine (40k+ images via ANN with Annoy and AlexNet) as MVP for Canadian trademark law firm
- Migrated Django app to GCP with zero infrastructure costs
- Developed TDD Django app with 83% test coverage
Integrated supplier API for platform handling 2k+ food delivery orders in under 2 months.
Designed REST endpoints; optimized API response 10x via SQL queries; refactored DB schema for consistency.
- Used Elasticsearch as read-replica for search performance
- Scaled scraping 100x with Scrapy and Scrapinghub
Built Human-in-the-Loop Active Learning workflow via CrowdFlower API; doubled F1-score for NER Logistic Regression with Uncertainty Sampling.
Operationalized multilingual NLP pipelines (Segmentation, Tokenization, POS, NER) for Romanian, Japanese, Chinese, German; contributed to EU-funded CAPER project (€5.6M).
Projects
Side projects and open-source work.
OpenOutreach
Open-source LinkedIn automation tool for agentic marketing. Python, Playwright, Docker, SQLite. Features AI message generation via LangChain with Jinja templates, multi-account management, A/B testing analytics.
SunnyPlans
AI-driven startup for solar land selection via geo-analytics and algorithms, utilizing PostGIS for spatial queries.
Chrome Extension for Stable Diffusion
Chrome Extension for generative AI prompt creation — 200+ installs in first month.
Web Scraping Projects
Built scraping pipelines for governmental RFPs (8 sites), subito.it, thefork.com, justeat.com, linkedin.com, admin.ch using Python, Scrapy, Playwright, Selenium.
Financial Dagster/dbt Lab
Open-source financial data pipelines with Dagster and dbt.
Financial Portfolio Management
Bayesian time-series forecasting for ETF portfolio — 40% ROI over 2 years.
Publications
Peer-reviewed IEEE conference papers.
CAPER: Crawling and Analysing Facebook for Intelligence Purposes
Focused on web scraping techniques for social media data collection and analysis.
Semantic Crawling: An Approach Based on Named Entity Recognition
Explored semantic enhancements to web crawling using NER for improved data visualization and extraction.
Community
Giving back through meetups and conferences.