Software Engineer
Software Engineer | AI/ML & Full Stack
Building LLM-powered systems and full-stack applications. Based near Hamburg and open to AI/ML roles.
Scroll to explore
PostgreSQL
Node.js
React
Tailwind CSS
JavaScript
Git
Python
PyTorch
FastAPI
LangChain
I build AI-powered systems and full-stack applications, from LLM pipelines to production web apps.
I'm a software engineer focused on AI/ML. During my internship at Geodouro I built a fully local RAG + Text-to-SQL assistant over 3 PostgreSQL databases with 500+ tables, taking answer accuracy from 13% to 77% with Qwen2.5 7B, hybrid retrieval and an LLM-as-Judge evaluation loop.
My projects span computer vision (CNN + ALPR for smart parking), LLM and voice systems, and full-stack platforms used by real teams. I pick the right tool for the problem: Python for AI/ML, React + Node.js for web, FastAPI for APIs.
I graduated in Computer Engineering from ESTGV/IPV in Portugal in June 2026 and now live in Schleswig, within commuting distance of Hamburg. I'm an EU citizen looking for a full-time Software Engineer (AI/ML) role. Meanwhile I'm building Zyllo, an LLM-driven SaaS for e-commerce, and learning German (A1).
Download Resume5+
Projects
10+
Technologies
2 years
Experience
Always
Learning
Feb 2026 – Jun 2026
Built a production RAG + Text-to-SQL assistant for SIGP, Geodouro's agricultural management platform. Replaced a free-roaming ReAct agent with a controlled pipeline (hybrid BM25 + MMR retrieval, schema pruning, dynamic few-shot, SQL validation with self-correction) over 3 PostgreSQL databases with 500+ tables. Accuracy on the company's golden set rose from 13% to 77%, with Qwen2.5 7B running fully locally and beating a 14B model.
Oct 2023 – Jun 2026
Graduated in June 2026 with a final grade of 15/20 (German 2.5). Final internship and project graded 20/20 (German 1.0).
A fully local assistant for Geodouro's agricultural management platform that answers questions from documentation (RAG) and from 3 PostgreSQL databases with 500+ tables (Text-to-SQL). A controlled pipeline with hybrid retrieval and schema pruning took accuracy from 13% to 77%.
Job seekers lacked realistic interview practice with instant feedback. Mockly generates role-specific questions with GPT-4o-mini, transcribes spoken answers in real time and scores them on clarity, depth, structure and relevance.
Manual parking monitoring was slow and error-prone. Built a real-time system using a CNN for spot occupancy and ALPR for license plate recognition, with ESP32-CAM gates, WebSocket-based live updates and automated event logging. Reduced manual checks to zero.
Developed a multi-platform training management application as a university project proposed by Softinsa (IBM). Built React and Flutter frontends with a Node.js REST API and PostgreSQL database. Managed onboarding workflows and real-time data synchronisation across platforms.
Since Jan 2026
Designing a SaaS platform that automates TikTok direct messages and follow-ups for e-commerce stores with an LLM-driven conversational engine. I defined the end-to-end architecture: database schema, auth, LLM orchestration and the TikTok API integration.