Versatile Swiss-Portuguese Junior Computer Scientist and AI/ML Engineer. Proven ability to design and deploy scalable architectures, utilizing LiteLLM and Docker to manage high-volume data retrieval and complex systems. …
Versatile Swiss-Portuguese Junior Computer Scientist and AI/ML Engineer. Proven ability to design and deploy scalable architectures, utilizing LiteLLM and Docker to manage high-volume data retrieval and complex systems. Proficient in LangChain/Agentic systems for automating workflows. Quick to adopt new technologies and focused on bridging advanced research with production-ready solutions. Currently completing an MSc thesis in Artificial Intelligence at USI.
• Designed and implemented an agentic system for financial analysis and reporting, capable of boosting analyst efficiency by up to 70% during testing phases.• Established core IT infrastructure by deploying containerized services (WebMail, MongoDB, hosting) on a secured VPS domain, ensuring self-managed security for sensitive financial data.• Developed targeted agentic workflows for web crawling and data analysis to generate comprehensive market research reports, informing robust financial plan creation.• Successfully applied analysis tools in a pilot project (gaming studio client) to drive actionable financial awareness and better client decision-making.
Designed and implemented an automated workflow for financial statement analysis and reporting using agentic systems.
Set up a VPS domain, deploying containerized services (WebMail, MongoDB, and website hosting).
Automated targeted web crawling and data analysis processes with agentic systems, streamlining data ingestion and reporting.
• Engineered a complex, distributed data pipeline infrastructure in two months to retrieve 5GB of longitudinal faculty data from over 100 universities, a scope impractical for manual collection.• Achieved high-performance data collection by utilizing LiteLLM to route traffic across 13 different LLM providers, circumventing API rate limits and completing retrieval in just 48 hours of compute time.• Designed and implemented an advanced AI agent system (LangGraph/LangChain), reducing data parsing errors from 18% to under 5% across large-scale, heterogeneous web sources.• Leveraged Wayback Machine API and Selenium for historical data retrieval, utilizing Supabase and PostgreSQL to manage storage and deliver the final processed data set.
Designed intelligent data pipelines for economic research, using LangGraph/LangChain, multi-provider LLM routing, and Supabase for structured storage.
Led prompt engineering and system oversight with LangSmith.
• Led the complete technical lifecycle (research, customization, and deployment) of the Siza Atlas platform as the sole IT resource, managing the architectural heritage of Álvaro Siza.• Successfully deployed the complex Arches Project framework, integrating services such as Elasticsearch on a private Ubuntu server while overcoming significant technical barriers.• Engineered a highly maintainable platform environment using Docker containerization.• Translated researcher data specifications to structure the digital archive for effective collaborative cataloging and long-term management.
Coordinated staffing and shift scheduling.
Supervised cooking staff and oversaw daily meal preparation as head chef.
Managed stocks.
• Directed all daily operational functions for a high-volume, 200-seat, two-kitchen restaurant over an eight-year tenure, overseeing financial planning, stock management, and quality control.• Led culinary operations as Head Chef, supervising a team of three assistants and optimizing kitchen/service workflows for smoother execution during rush hours.• Cultivated a high-performance team culture by achieving strong staff retention over several years through organizational clarity and fair scheduling.
• Mentored and guided 23 undergraduate students through core database concepts, relational schema design and advanced SQL querying.• Specialized in teaching data modeling using UML Class Diagrams, successfully guiding students through the translation of over 8 conceptual-to-relational models.• Customized course material delivery utilizing analogies and repetition, leading to 85% of students reporting high confidence in complex database principles.• Instructed practical application across three major full-stack assignments, covering advanced SQL querying, PHP integration, and MySQL database management.
Taught core database concepts and skills, including relational schema design, UML class diagrams, conceptual-to-relational model mappings, SQL, PHP, and MySQL.
Provided personalized and group math tutoring to high school students, aligned with national testing standards.
Designed targeted lesson plans, practice exercises, and review sessions to strengthen students’ problem-solving skills and test-taking confidence, resulting in notable improvements in their exam performance.
Led the research, customization, and deployment of the Siza Atlas collaborative platform for cataloging and managing the cultural heritage of Álvaro Siza’s work using the Arches Project framework.
Implemented and containerized the solution on ISTAR’s private Ubuntu server, ensuring a secure and scalable environment.
Information Processing course⦁ Developed a Flutter application that modelled team productivity using stochastic simulations.⦁ Based the model on a strong mathematical foundation, incorporating 11 key lifestyle and resource factors to generate predictive insights.⦁ Provided analysis for performance optimization and resource planning in organizational settings.
Pilot Project with Hurricane Studios⦁ Built a custom web-scraping pipeline that collected high-volume market data from 50+ sources, successfully ingesting over 50,000 user feedback entries.⦁ Integrated NLP and AI models for sentiment analysis, achieving a comprehensive, meaningful conclusion across the entire dataset within 30 minutes.⦁ Delivered interactive dashboards enabling the client to make data-driven decisions based on real-time sentiment analysis and trend detection.
⦁ Developed and deployed an agent-based system to ingest financial statements and automatically generate Swiss GAAP FER compliance reports.⦁ Reduced the manual effort required for compliance verification and audit streamlining by approximately 60%.⦁ Incorporated advanced data parsing and validation techniques to minimize error rates and ensure high audit validation accuracy.
Deep Learning course⦁ Pre-processed 200,000 dialog pairs from the Cornell Movie Dialog dataset and trained a custom Transformer model using PyTorch.⦁ Designed a 10–12 layer Transformer architecture (trained locally) that achieved more coherent, context-aware responses than standard Seq2Seq approaches.⦁ Showcased advanced natural language understanding (NLU) capabilities and expertise in modern deep learning model development.
Information Modeling & Analysis course⦁ Leveraged unsupervised learning techniques (trained on the full TensorFlow library) to automatically identify and decompose large, monolithic Java classes.⦁ Improved maintainability and OOP compliance by increasing code cohesion and reducing coupling by 35% across the refactored code base.⦁ Automated the refactoring process, enabling a transition toward more modular, maintainable architectures.
Computer Vision & Pattern Recognition course⦁ Created a video pre-processing pipeline combining classical Computer Vision (CV) techniques (Canny, Hough, DLT) to reconstruct camera perspectives in snooker footage.⦁ Achieved an estimated 98% accuracy in projecting the table and ball movements onto a top-down view, significantly enhancing shot analysis and angle measurement capabilities.⦁ Informed more accurate performance feedback by enabling clear visual analysis, despite initial video distortions.
High Performance Computing course⦁ Implemented domain decomposition and parallel computing techniques (MPI) to solve non-linear partial differential equations in C++.⦁ Demonstrated high efficiency and distributed computing expertise by achieving near-linear speedup up to 12 cores.⦁ Built a scalable system foundational for large-scale scientific simulations.
Distributed Systems course⦁ Designed a distributed solution for real-time data ingestion of MQTT stream data from a simulated mice lab.⦁ Managed real-time ingestion of over 1,000 MQTT data streams per minute into MongoDB and MariaDB.⦁ Enabled continuous data capture, monitoring, and scalable analysis, managed via a Flutter frontend, for experimental research.
The AI-first Recruitment Operating System that unifies and automates the entire hiring lifecycle for modern enterprises worldwide.
Dr Job FZ LLC. 2026 © All Rights Reserved
