Software Engineer focused on Backend Systems, Distributed Computing and Applied AI.
M.Sc. Computer Science and Engineering @ University of Bologna
Building reliable software systems and experimenting with LLMs for code generation.
Fine-tuned small language models to generate executable Akka/Scala code using parameter-efficient fine-tuning.
Highlights
- Fine-tuned Llama 3.2 and Qwen 2.5 using LoRA / PEFT
- Built automated dataset preparation and code validation pipelines
- Evaluated generated code using executability and code-quality metrics
- Increased framework-specific executable code generation substantially compared with the base models
Stack: Python · PyTorch · Hugging Face · LoRA · Scala · Akka
Interested in designing reliable systems involving:
- distributed services
- asynchronous messaging
- concurrency
- event-driven architectures
- APIs and databases
Core stack
Java Spring Boot Kafka Scala Python Docker SQL
Automation tool for organizing large media collections inside Telegram forum topics.
- Built an asynchronous Telegram bot in Python
- Automatically routes media to topics based on filename metadata
- Creates and persists topic mappings using SQLite
- Uses per-topic async locks to prevent duplicate topic creation during concurrent batches
- Handles videos, images, audio and documents with normalized captions and filenames
- Detects stale Telegram thread mappings and automatically recreates/retries them
- Includes fallback handling for failed uploads and message routing
Python AsyncIO Telegram Bot API SQLite Automation Media Processing
- LLM inference and fine-tuning
- Retrieval-Augmented Generation
- Distributed systems
- Reliable backend architectures
- AI-assisted developer tooling