Luca Corvitto

I'm an NLP

About

I'm an NLP Engineer at Babelscape ®, where I focus on designing, developing and optimizing NLP and LLM agentic systems.

Previously, I worked as as an AI Research Engineer at CAELUS, on GenAI applications for the nuclear sector, focusing on explainability, safety, and digital innovation, contributing to the broader mission of decarbonization and Net Zero.

My expertise lays in designing LLM agentic systems using LangGraph, LLM inference APIs, knowledge graphs and vector databases.

I'm also pursuing a Master's degree in Artificial Intelligence and Robotics at Sapienza University of Rome. My academic journey includes research in generative AI for forensic applications and audio-based AI for healthcare, with two peer-reviewed publications.

Driven by a strong ethical commitment, I believe in the potential of AI to promote sustainability, social equity, and animal welfare. Besides this, I am actively involved in advocacy for animal rights and environmental sustainability.

In my free time, I explore new technologies through personal projects and stay engaged in learning.

Computer, System & AI engineer.

I am currently working on my Master's Degree, in which I addressed different topics in AI and robotics, such as:

  • Deep Learning: CNNs, RNNs, Transformers, Graph NNs
  • Computer Vision: Face Recognition, Object Detection, Image Segmentation, GANs, Deepfake detection
  • Natural Language Processing: Text Classification, SRL, NER, NLU and NLG
  • Neural Networks: Non-Linear Modeling, Gated RRN, VAE, Multimodal Learning, Diffusion Models
  • Reinforcement Learning: Policy Iteration, Q-Learning, Sarsa, DQN, DDQN
  • Human Robot Interaction: Multi-modal interaction, Robot Speech Understanding, Visual Perception
  • Autonomous and Mobile Robotics: Wheeled Mobile Robots, Perception, Localization, Motion Planning
  • Large Language Model Agents: Reasoning, Tool use, LLM Agent Infrastructure, Agentic RAG, Human-Agent Interaction, Multi-Agent Collaboration

Resume

Work Experiences

NLP & LLM Engineer

2025 - Present

Babelscape ®, Rome, Italy

AI Research Engineer

2024 - 2025

CAELUS S.R.L., Rome, Italy

  • Designed and implemented an agentic RAG system using LangGraph, both in local/private configurations (with Ollama models and a local Qdrant client) and via public APIs (e.g., OpenAI's GPT APIs with vector search over Pinecone).
  • Built training pipelines for LLMs, including instruction tuning, supervised fine-tuning, and reinforcement learning techniques.
  • Conducted research on state-of-the-art approaches for RAG, agentic LLMs, and the application of GenAI in regulated industries.
  • Collaborated cross-functionally with IT and Cybersecurity teams to deploy secure end-to-end pipelines that prevent private data leakage.
  • Participated in international conferences and expos (e.g., NuclearEurope, AFNBP, NPE) to present the product and engage with industry professionals.

English Tutor

2023

Sapienza University of Rome, Italy

  • Professor Assistant: Assisting the professor in carrying lectures out.
  • Student Tutor: Assisting students in the preparation of the exam.

Adidas Store Employee

2017

Adidas Official Store, Rome, Italy

  • Shop Assistant: Assisting customers in the purchasing process, both in Italian and English.

Education

Master's Degree in Artificial Intelligence and Robotics

2021 - Present

Sapienza University of Rome, Italy

Second-level degree, in which the knowledge of Artificial Intelligence and Robotics is introduced and deepened with both theoretical courses and practical projects.

Bachelor's Degree in Computer and System Engineering

2018 - 2021

Sapienza University of Rome, Italy

First-level degree, in which, in addition to the basic knowledge of mathematical analysis, physics and statistics, it introduces to the concepts of computer programming, electronics, systems theory and automation.

Diploma in Scientific Lyceum

2013 - 2018

Liceo Statale A. Meucci, Aprilia, Italy

An high school that focuses more on maths-based courses like mathematics (linear algebra and geometry), physics and chemistry. It has also a course on the latin language and literature.

Skills

Languages

  • Italian, English

Programming & Markup languages

  • Python, Java, JavaScript, C, C++, Matlab, HTML, CSS, PHP, LaTeX

Computer Science libraries & tools

  • Windows, Linux, Git, Github, Docker, Pandas, Numpy, SciPy, Matplotlib, Bootstrap, Streamlit...

ML libraries & platforms

  • Pytorch, Lightning AI, Weights & Biases, LangChain and LangGraph, AutoGen, Pinecone, Qdrant, Google Colab, Scikit-learn, OpenCV, Huggingface, Transformers, NLTK, SpaCy...

Publications

Speech and Language Impairment Detection by Means of AI-Driven Audio-Based Techniques

29/07/2024

Proceedings of the 9th International Conference of Yearly Reports on Informatics, Mathematics, and Engineering (ICYRIME 2024)

On the use of Stable Diffusion for creating realistic faces: from generation to detection

26/06/2023

IEEE International Workshop on Biometrics and Forensics (IWBF)

Certifications

Advanced Large Language Model Agents MOOC

07/2025

Berkeley RDI

Introduction to LangGraph

04/2025

LangChain Academy

Large Language Model Agents MOOC

01/2025

Berkeley RDI

Portfolio

  • All
  • Projects
  • Papers

Autogen LLM agents system

A LLM agents system, built with Autogen (AG2), that produces a summarized review of a specified restaurant out of a list of several restaurant reviews.

AIN-RAG streamlit web app

An auto evaluative RAG chatbot extracting information about AI regulations in Europe and United States.

Speech and Language Impairment Detection by Means of AI-Driven Audio-Based Techniques

AI-based tool for detecting Speech and Language Impairments (SLI) using audio signals, achieving 96% accuracy with advanced feature extraction and data augmentation techniques.

DRDQL for Atari

A pytorch implementation of Deep Recurrent (Double) Q-Learning for the Atari environment with Gymnasium.

On the use of Stable Diffusion for creating realistic faces: from generation to detection

An analysis on the generation power of Stable Diffusion for creating photorealistic images and on the ability of Deep Learning models to recognize them.

GANime

GAN architecture trained to generate anime-like faces.

Natural Language Generation in the style of Asimov Isaac (NLGAI)

GPT-2 model trained on a dataset composed of a huge corpus of Isaac Asimov's books to generate text in the style of the famous sci-fi writer.

Emotional Responsive Interaction with Pepper

Human Robot Interaction performed with a Pepper robot showing emotive reactions to the opinions of the user while collecting personal data.

IRRT* python implementation

Improving the performance of RRT planners using Informed Sampling. Python implementation both in Euclidean and Kinodynamic domains.

Tutoring

In Presence

Contact me and we will procede to a standard face-to-face tutoring session.

From Remote

Contact me and we will meet online on the meeting platform you prefere to procede to a remote tutoring session.

Contact

Location

Aprilia (LT), Italy, 04011

Email me

luca99corvitto@gmail.com

Connect with me

linkedin.com/in/lucacorvitto

Follow me

github.com/LucaCorvitto

Write me

@LucaCorvitto