Riccardo Zuliani

Machine Learning Engineer

Riccardo Zuliani

I take data-driven solutions from prototype to production: data pipelines, model training, LLM-based agents and RAG assistants, and deployed web and mobile applications.

View projectsGet in touchDownload CV

About

Research mindset, industrial focus.

Machine Learning Engineer with a solid academic background and hands-on experience in deep learning, time-series anomaly detection, computer vision, and LLM-based agents. Skilled in Python, PyTorch, LangChain, and full-stack development, taking data-driven solutions from prototype to production: data pipelines, model training, RAG assistants, and deployed web and mobile applications.

I am proactive, research-oriented, and passionate about AI in industrial contexts.

Focus
Deep learning, time-series anomaly detection, computer vision, LLM agents
Current role
Freelance Software Engineer, The Italian Elixir & Training Hub
Education
MSc Computer Science, Ca’ Foscari University of Venice
Portrait of Riccardo Zuliani in a blue suit and a navy patterned tie

Experience

From raw data to shipped products.

  1. Present

    Remote

    Freelance Software Engineer

    The Italian Elixir (THIE) & Training Hub

    • THIE Business Engines. Designed and built a suite of independent, test-driven TypeScript services for a hospitality business: quoting, booking operations, ads management, reviews and property library.
    • THIE Architecture. Defined domain boundaries, idempotent commands and optimistic concurrency, with provider-neutral adapters over Google Sheets, Apps Script and Wix.
    • THIE Operator Console. Delivered an internal console for staff to calculate and revise quotes, developed locally and published only after visual approval.
    • Training Hub. Built a coaching SaaS for strength coaches and athletes with Flask, PostgreSQL and htmx, shipped as a PWA and Capacitor iOS/Android app with offline logging, Stripe subscriptions, two-factor authentication, GDPR consent and an opt-in AI assistant.
    • TypeScript
    • Flask
    • PostgreSQL
    • htmx
    • PWA
    • Capacitor
    • Stripe

  2. Bologna, Italy

    Machine Learning Engineer & Researcher

    NAIS Engineering Srl

    • Data Analysis. Developed multiple web applications with Dash Plotly to visualize project results and analyses.
    • Data Engineering. Built multiple aggregated datasets on Formula 1 engines at different levels of depth, with one record sampled every 0.005 seconds (200 Hz).
    • Neural Network Modeling. Developed and evaluated different anomaly detection models on time-series datasets.
    • SCADA Dashboard Development. Developed a SCADA dashboard from scratch using Ignition Perspective to automate PLC data collection, enabling downstream ML model development and anomaly detection pipelines.
    • Anomaly detection
    • Time-series
    • Dash Plotly
    • Ignition Perspective

  3. Venice, Italy

    Probability and Statistics Tutor

    Ca’ Foscari University of Venice

    • Tutorials. Delivered 10 classes and assisted 30 students with probability and statistics problems, by hand and in R, including one-on-one support.
    • Probability
    • Statistics
    • R

Projects

Selected projects.

  • Master’s thesis

    Active Learning with Dynamical System

    A novel Active Learning methodology that uses the Graph Transduction Algorithm to select the most difficult samples from a subset and add them to the known set.

    • Active learning
    • Graph transduction
    • Dynamical systems

  • Full-stack · RAG agent

    Molino Bertolo: Website, Shop and AI Assistant

    A Next.js 16 and PostgreSQL storefront (235 static pages, admin dashboard, cart) with a ReAct RAG agent over 307 technical sheets and recipes (Pinecone, OpenAI) that recommends products and adds them to the cart.

    • Next.js
    • PostgreSQL
    • RAG
    • ReAct agent
    • Pinecone
    • OpenAI

  • LLM agent

    Strudel Assistant

    A documentation assistant for the Strudel live-coding language: hybrid BM25 and Pinecone retrieval, a ReAct agent that validates patterns in a Node sandbox, and melody transcription from humming with librosa.

    • Hybrid retrieval
    • Pinecone
    • ReAct agent
    • librosa

  • Computer vision

    Silhouette-Based Space Carving

    Reconstructs the 3D shape of an object by capturing it from multiple angles during rotation and progressively removing empty space based on its silhouettes.

    • 3D reconstruction
    • Computer vision

Education

Academic background.

  • Master of Science in Computer Science

    Data management and analysis

    Ca’ Foscari University of Venice, Italy

    Thesis
    Active Learning with Dynamical System
    Relevant courses
    • Artificial Intelligence
    • 3D Geometric Computer Vision
    • Statistical Inference

  • Bachelor of Science in Computer Science

    Data Science

    Ca’ Foscari University of Venice, Italy

    Thesis
    Dash AutoML Benchmark
    Relevant courses
    • Algorithms and data structures
    • Databases
    • Data Analysis
    • Predictive Analytics

Skills

Tools of the trade.

Programming languages

  • Python
  • TypeScript
  • C++
  • SQL
  • R

Frameworks & libraries

  • PyTorch
  • OpenCV
  • Scikit-learn
  • Hugging Face
  • LangChain
  • LangGraph
  • Pandas
  • NumPy
  • Matplotlib
  • Seaborn
  • Dash Plotly
  • Flask
  • FastAPI
  • Next.js

Technologies

  • Deep Learning
  • Machine Learning
  • Computer Vision
  • LLM Agents
  • RAG
  • PostgreSQL
  • Pinecone
  • Git
  • Linux
  • Docker
  • AWS
  • Ignition

Spoken languages

  • Italian (native)
  • English (B2+, upper-intermediate)

Contact

Let’s talk.

Interested in machine learning, anomaly detection, computer vision, LLM agents or industrial AI? The best way to reach me is by email.