Daniele Moltisanti

I'm

About Me

Daniele Moltisanti

Data Scientist & Machine Learning Engineer

A results-driven Data Scientist with a strong technical background, experienced in leading successful data-driven projects and high-performing teams. Specializing in leveraging Generative AI to deliver innovative solutions. Passionate about spreading AI culture through my initiative stAI tuned.

  • Email: danielemoltisanti@gmail.com
  • Contact: +39 3337337311
  • LinkedIn: Profile
  • Latest Role: Lead Data Scientist at Sky Italia
  • Skills Focus: Machine Learning, Deep Learning, NLP

Awards

  • Nova 111 List (Italy) 2022:Recognized as a top talent in Entertainment, Media & Communications among professionals under 35, highlighting my contributions and impact in the field.
    • Nova 111 2022
  • Nova Talent Member:Merit-based access network where the top 3% of global talents connect with each others .
    • Membership Certification
Daniele Moltisanti

Skills

NLP & Generative AI

NLTK | Spacy | Gensim
  • Sky Italia: Clustering Analysis on conversations
  • Sky Italia: Topic Modeling algorithm
  • Youtility Center: Clustering Analysis on chatbot conversations
Langchain
  • Sky Italia: Q&A Agent
  • Sky Italia: GenAI Hackathon organised by Sky & Google
Hugging Face
  • Sky Italia: Q&A Agent
  • Sky Italia: Use of OpenSource Embedding and LLMs
LLM
  • Sky Italia: OpenSource Embeddings
  • Sky Italia: OpenSource LLMs

Programming Languages & Frameworks

Python
  • From University on, I used mainly Python
Pytorch
  • Sky Italia: Propensity Models & Synthetic Data Generation
  • Coach for Pytorch course
  • Master Thesis: Anomaly Detection on Images
  • Personal Project: Financial Anomaly Detection & Forecasting
  • Personal Project: Genre Detection
  • Personal Project: Soccer Match Prediction
Tensorflow / Keras
  • Accenture: Custom NER algorithm
Scikit-Learn
  • Sky Italia: Models for classification/regression tasks
SQL
  • Sky Italia: ETL for the ML Processing
  • Sky Italia: Analysis on Data

Cloud Platforms

Google Cloud Platform
  • Sky Italia
  • Accenture
Amazon Web Services
  • Youtility Center

Development Tools

Docker & Docker-Compose
  • Youtility Center: Test Deployment of Chatbot ecosystem
Apache Airflow
  • Sky Italia: Deployment & Scheduling of ML pipeline for ETL and model training/prediction
Git
  • All project I did were versioned using Git

Soft Skills

Leadership
  • stAI tuned: led projects and more than 10 people
  • Sky Italia: led data science projects
  • Youtility Center: led the end-to-end chatbot pipeline and people working on that
Problem-Solving
  • Every time I need to find a solution to a specific problem
Communication
  • With Technical People
  • With Non-Technical People, explaining the value of the data science model outcome
Organization
  • I base each activity on a strong organization

Languages

Italian
  • Native
English
  • Professional level
  • B2 Level Certification
  • University & work in english
Spanish
  • Professional Level
  • Lived 6 months in Valencia, Spain
French
  • Basic knowledge
  • A1 Level

Expertise

Here you find the expertise from my work experiences

Project Image Description

Generative AI


  • Q&A Agent: developed a conversational assistant, trained on a specific domain, to answer on specific question related to the given domain. Used langchain and a multi-LLM orchestration through the langchain tools.

  • Prompt Engineering: the strong use of GenAI tools make me understand the effective and efficient ways to take advantages from the LLMs. Able to refine prompts for reaching correct answers from LLMs. Experienced with the most known LLM tool like GPT, Gemini, but also with open-source LLMs, from Hugging Face, using LlamaCpp

  • Synthetic Data Generation: developed a Diffusion Model able to generate synthetic data from tabular datasets, taking care of the privacy and utility of those synthetic data. This model was created from scratch in Pytorch

  • Participation to GenAI Hackathon: enrolled in a GenAI hackathon organised by Sky Group and Google, implementing one business use case using GenAI, using Langchain framework and Palm as LLM

    • Predictive Analytics


  • Propensity Models: developed models for predicting the customer behaviour, contributing to improve the customer lifetime value. Used Supervised and Semi-Supervised Learning models.

  • Time-Series Analysis: conducted now-casting of financial time-series data to detect anomalies and identify trends, aiding in data-driven decision-making processes

    • Project Image Description
    Project Image Description

    Advanced NLP Analytics


  • Topic Modeling: developed various models that assign the topic to a given text, working both on short texts and long texts, using Semi-Supervised and Unsupervised Learning Approaches.

  • Clustering Analysis: working on producing analysis basing on the sentence meanings, using word and sentence embeddings pre-trained

  • Virtual Assistant: developed various custom virtual assistant in python, using the Rasa framework, basing on specific domain Took care of the architecture and deployment phases, implementing a asynchronous server that handled and orchestrated various chatbot.

  • Named Entity Recognition (NER): developed a custom NER for extracting entities automatically. The NER model was based on char and word embeddings trained and fine-tuned on the available dataset

    • Computer Vision


  • Image Denoising: worked on denoising document ID image to improve the OCR performance in detect the text

  • Anomaly Detection: detect anomalies on images, using an research approach of processing the image in a patch-wise way

  • Background Subtraction: implemented a Gaussian mixture model on highway videos for removing the cars on the foreground

    • Project Image Description

    Initiatives

    Here are some of the projects I've worked on, related to my AI passion.

    Project Image Description

    stAI Tuned

    Open Knowledge Community where each member shares his knowledge and research bites related to an AI topic in the real life.

    Goal: spreading AI insights and the AI culture from practical applications to courious news. stAI tuned is live on the blog website, Linkedin (addressed to an expert audience) Instagram and Tiktok (addressed to a newbie audience) accounts.

    Achievements:
  • 1300+ follower on Linkedin
  • 500+ follower Instagram
  • Up to 14k views and 8% of engagement rate in last 12 posts on Instagram
  • Blog: 35+ articles | 250+ active users | 600+ sessions
    • My Articles stAI tuned | Blog Website Linkedin
    Project Image Description

    Nova Mentor

    Mentoring university students in their growth, defining their own career path

  • 4 mentored students
  • Mentored italian and spanish students
    • Mentor Certification

    AI Coach | Profession AI

    As a Coach I review and help students of ProfessionAI Master Courses in writing the Machine Learning solution to a given problem

  • 150+ Projects reviewed
  • 3 type of ML Projects
  • Support during development of the solution, if needed
    • Project Image Description
    Project Image Description

    Deep Learning Course | Pytorch

    7 practical hours of Deep Learning course where learning:

  • what is an artificial neural network
  • how to create it in pytorch
  • practical lab sessions
    • Course Github Lab Sessions

    Personal Projects

    Here are some personal projects developed in my free time and my university career

    • All
    • Computer Vision
    • Time Series Forecasting
    • NLP
    • Finance

    Resume

    Professional Experience

    Lead Data Scientist

    03/2022 - Present

    Sky Italia, Milan, Italy

    • Led technically small project teams of data scientists (up to 2 data scientist), providing both mentorship and hands-on support in daily tasks and career growth
    • Actively collaborated with cross-functional teams implementing Machine Learning solutions for reducing the churn rate up to 30%
    • Developed and executed a technical strategy for automatizing manual business processes for the continuous improvement of the Virtual Agent, using ML and NLP, enhancing operational efficiency by 50%
    • Implemented a topic modeling algorithm on conversational text within the Virtual Agent channel, achieving a 40% reduction in conversation analysis time
    • Introduced and experimented Generative AI techniques to create a smart data assistant, significantly enhancing data accessibility, visualization, and decision-making processes for Data Employees, leading to a 40% reduction in data retrieval times.

    Tools: Python, Pytorch, SQL, GCP, Airflow, Generative AI, Langchain, LLM

    R&D Data Scientist

    01/2021 - 03/2022

    Youtility Center (Gruppo Distribuzione SPA,. Milan, Italy

    • Managing Technical aspects of Conversational AI projects (chatbots).
    • Achieved over 95% accuracy relative to NLU (natural language understanding) models.
    • Achieved over 95% accuracy relative to NLG (natural language generation) models.
    • Handled complex chatbot (more then 150 intents).
    • Developing and managing the backend and the infrastructure needed for the conversational models
    • Developed multi-bot infrastructure, managing multiple bots inference in parallel.
    • Developed multi-language and multi-domain chatbots.
    • Making use of AWS services, such as storage (S3), IaaS (EC2) and database (Aurora).

    Tools: Python, Tensorflow 2.x, AWS, Rasa 2.x, Docker, Docker-Compose, NLTK, Spacy, SQL

    Data Scientist

    05/2020 - 01/2021

    Accenture S.p.A., Milan, Italy

    • Developed a custom NER model for the document understanding task related to Financial Industry, starting from a small dataset.
    • Achieved over 95% accuracy relative to the entity classification of documents, even if the small input data.
    • R&D relative to NLP tasks such as Entity Extraction and Named Entity Recognition (NER).
    • Developed a custom Computer Vision denoising model for noise reduction in documents, achieving relevant improvements for the OCR task.
    • Developed projects making use of Google Cloud Platform (GCP) services, such as IaaS (compute instances), PaaS (app engine), storage

    Tools: Python, Pytorch, Tensorflow 2.x, Keras, CUDA, Git, OpenCV, GCP

    AI engineer - Internship

    09/2018 - 02/2019

    Blue reply, Milan, Italy

    • Contributed in the development of a virtual assistant, where I developed the cognitive model for classifying fashion items in images.
    • Experienced with IBM Watson services.
    • Integrated the cognitive model into the telegram channel, handling RestAPI.

    Tools: Python, Pytorch, CUDA, Git, IBM Watson

    Education

    Master's Thesis Researcher

    2019 - 2020

    Politecnico di Milano, Milan, Italy

    Title: "Anomaly Detection through Ganomaly: patch-wise training and transfer learning"

    Goals: introducing improvements to the Ganomaly state-of-the-art for Anomaly Detection in order to achieve a more efficient training for any-dimension images and a more effective performances through the transfer learning technique.

    Master's Degree in Computer Science and Engineering

    2017 - 2020

    Politecnico di Milano, Milan, Italy

    Artificial Intelligence path

    Athens Programme - Intensive University Cource - From Complexity to Intelligence

    2019

    Telecom Paris-Tech, Paris, France

    Participation to the European Athens Programme, attending an intensive course related to the complexity aspect with respect to the artificial intelligence algorithms

    Erasmus+ project at Universitat Politecnica de Valencia

    2018

    Universitat Politecnica de Valencia, Valencia, Spain

    Bachelor's Degree in Computer Engineering

    2014 - 2017

    Politecnico di Milano, Milan, Italy

    Languages

    Italian

    • Native

    English

    • B2 Level - TOEIC | ETS Global B.V.

    Spanish

    • B2 Level - FIDESCU

    Certifications

    Associate Cloud Engineer

    See Credential

    Google Cloud Platform - 09/2020

    Contact

    Location:

    Milan, Italy