Skip to content
View AlessandroS01's full-sized avatar

Highlights

  • Pro

Block or report AlessandroS01

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.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. 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
AlessandroS01/README.md

Hi there πŸ‘‹, I'm Alessandro Seghini

πŸ‘¨β€πŸ’» About Me

  • πŸŽ“ Background: M.Sc. in Artificial Intelligence Engineering from UniversitΓ€t Passau and B.Sc. in Computer Science from UniversitΓ  Politecnica delle Marche.
  • πŸ› οΈ Engineering Mindset: Strong advocate for clean architecture, Domain-Driven Design (DDD), static typing, and automated testing. I prefer deterministic code, strict schemas, and zero-ORM overhead over fragile layers.
  • πŸ”­ Current Focus: Developing Strataβ€”an air-gapped, offline hybrid RAG knowledge engine built with Python, SQLite WAL, and local vector reranking.
  • 🏎️ Leadership: Former Software Engineering Team Lead for the Polimarche Racing Team, architecting telemetry backends for 50+ engineers.
  • πŸ’¬ Ask Me About: Python backend design, high-frequency telemetry ingestion, streaming data reliability, local RAG architectures, and Spring Boot testing patterns.

🧰 Tech Stack & Tooling

Languages Python Java SQL Dart TypeScript Kotlin
Backend & APIs FastAPI Spring Boot Pydantic RESTful APIs DDD
Data & Storage SQLite WAL Qdrant MySQL H2 Firebase Streaming
AI & Retrieval PyTorch LangChain Ollama Hybrid RAG LLM Evaluation
Tooling & Quality Docker CI/CD JUnit5 Mockito Mypy Ruff

πŸš€ Highlighted Projects

πŸ” Strata

Air-gapped, offline hybrid RAG engine and local personal knowledge base.

  • Architected via Domain-Driven Design and SQLite in WAL mode, delivering sub-5ms local reads without ORM overhead.
  • Implemented an AST-aware Markdown chunker generating Small-to-Big relationships with SHA-256 caching to skip duplicate vector embedding steps.
  • Dual-channel retrieval: dense vector search (Qdrant) + sparse lexical search (BM25) fused via Reciprocal Rank Fusion (RRF) and cross-encoder reranking.

Python β€’ SQLite (WAL) β€’ Qdrant β€’ FastEmbed β€’ rank_bm25 β€’ Mypy

High-integrity multi-modal telemetry pipeline & automated LLM auditing suite.

  • Engineered end-to-end Python pipeline processing streaming 100 Hz PPG/ACC data into structured clinical payloads.
  • Resolved Bluetooth jitter with a 40 Hz interpolation protocol, yielding 100% deterministic reliability across validation runs.
  • Created a scalable LLM-as-a-Judge evaluation framework assessing model decision boundary stability, safety, and latency.

Python β€’ Pydantic β€’ LangChain β€’ PyTorch β€’ Ollama

Scalable telemetry management platform for a Formula SAE team of 50+ members.

  • Designed a modular Spring Boot REST backend applying Service-Repository patterns and Role-Based Access Control (RBAC).
  • Guaranteed regression stability through comprehensive unit and integration suites using JUnit5, Mockito, and H2.
  • Shipped a cross-platform Flutter client centralizing dynamic vehicle setup configurations and test runs.

Java β€’ Spring Boot β€’ JUnit5 β€’ Flutter β€’ Dart β€’ Firebase

Real-time edge control and computer vision system on Raspberry Pi 4.

  • Built an autonomous closed-loop pipeline translating real-time computer vision metrics directly into motor commands.
  • Trained and deployed edge vision models (YOLO) with noise/vibration augmentations, reaching 0.992 mAP50 for edge line and object tracking.

Python β€’ YOLO β€’ Roboflow β€’ Raspberry Pi 4 β€’ Computer Vision


πŸ“Š GitHub Activity & Metrics

GitHub Stats Top Languages
GitHub Streak

πŸ“¬ Let's Connect & Collaborate

I am always open to discussing backend architectures, local AI retrieval systems, or software engineering opportunities.

LinkedIn Email GitHub

Footer Banner

Pinned Loading

  1. Strata Strata Public

    Python

  2. Callisia-WearLM Callisia-WearLM Public

    Combining traditional Machine Learning and Large Language Models for Healthcare: A Hybrid Approach for Interpreting Wearable Data

    Jupyter Notebook

  3. thesis-polimarche thesis-polimarche Public

    Dart

  4. PiCarX-AI-Vision-Control PiCarX-AI-Vision-Control Public

    Python 1

  5. ladolcevita ladolcevita Public

    TypeScript

  6. home-energy-tracker home-energy-tracker Public

    Java