Skip to content
View bGuzzo's full-sized avatar
🤓
🤓
4 followers · 1 following

Block or report bGuzzo

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
bGuzzo/README.md

Hi, I'm Bruno Guzzo 👋

💼 Professional Overview

I am a Systems Engineer at AWS (Dublin), working within the Managed Operations team since April 2026.

🎯 Interests & Focus

I am passionate about Linux, Cloud, SWE, DevOps, AI/ML (NLP, Anomaly Detection, Time Series), and Finance. I use this profile to share personal projects and ideas; feel free to reach out for collaboration.

🚀 Projects

🛠️ Personal Projects

  1. [In Progress] - Fin-FFB: A finance-specialized BERT-like embedding model designed to embed large documents on consumer-grade hardware. It leverages recent NLP advancements to implement a fat and shallow architecture, bypassing the context size limitations of common models like Mini-LM.
  2. [Coming Soon] - Fin-PT: A financial pre-trained transformer model designed to apply GPT architectures to predict the next OHLC data point for a given ticker history.

🎓 Academic Projects - UNICAL/DIMES AI/ML Engineering MSc

  1. Italian-LLaMA-Project: Fine-tuned the 8B LLaMAntino-3 model for Italian proficiency using 4-bit quantization and LoRA. Implemented a complete RAG chatbot system using LangChain and FAISS.
  2. MSc Thesis: Semi-Supervised Anomaly Detection: A rigorous experimental analysis of undersampling techniques (e.g., ENN, NCR, and Tomek Links) to improve the performance and efficiency of Auto-Encoders for semi-supervised anomaly detection.
  3. Transformer-GNN for Link Prediction: Designed and built a custom, deep Graph Neural Network (GNN) inspired by the Transformer architecture. Trained from scratch on a custom 7-million-node dataset (crawled from Wikipedia) for semantic link prediction.
  4. Anomaly-Transformer-Analysis: A deep-dive implementation and comparative analysis of the "Anomaly Transformer" paper, benchmarking its novel Anomaly-Attention mechanism against standard self-attention for time-series anomaly detection.

🧰 Core Competencies

Category Technologies
AI & ML PyTorch, LangChain, LLM Fine-Tuning (LoRA), RAG, NLP, Pandas, Vertex AI (Gemini)
Cloud & DevOps AWS, CDK, Google Cloud (GCP), Terraform, Docker, CI/CD (GitHub, GitLab), Cloud Run, BigQuery
Software Engineering Python (Flask), Java (Spring Boot), JavaScript (Angular), SQL & NoSQL

📫 Get In Touch

Pinned Loading

  1. MSc-AI-ML-thesis-anomaly-detection MSc-AI-ML-thesis-anomaly-detection Public

    An experimental analysis of under-sampling in deep semi-supervised anomaly detection. This MSc thesis evaluates how different sampling strategies affect the performance and efficiency of the AE-SAD…

    TeX

  2. Anomaly-Transformer-Analysis Anomaly-Transformer-Analysis Public

    An in-depth analysis of the Anomaly Transformer, with experiments on hyperparameter tuning, optimizers, and a head-to-head comparison with a standard Transformer Encoder.

    Python

  3. transformer-gnn-link-prediction transformer-gnn-link-prediction Public

    A deep, transformer-inspired Graph Attention Network (GAT) for link prediction on semantic graphs. This project features a custom GNN architecture and a large-scale dataset built from Wikipedia.

    Python

  4. Italian-LLaMA-Project Italian-LLaMA-Project Public

    An MSc NLP project on fine-tuning LLaMAntino-3-8B for Italian. Explores efficient training on consumer hardware using PEFT/LoRA and includes extensive evaluation (Perplexity, INVALSI) and chatbot/R…

    TeX