Jul 2025 – Oct 2025 · 3 mos · Madrid, Community of Madrid, Spain

• Developing AI agents for root cause analysis in image recommendation systems, enabling automatic detection of incorrect images and explanation of recommendation decisions.

Sep 2020 – Jun 2025 · 4 yrs 9 mos · Madrid, Community of Madrid, Spain

• As a Machine Learning Engineer at INTech, I built ML and GenAI systems to enhance visual customer experiences on Amazon’s shopping website. Previously, I worked in the Books/Kindle organization, where I developed ML solutions to improve book metadata quality. Main achievements:
• Led the MLOps automation roadmap for an Applied Science team, cutting model evaluation time from 3 weeks to under 5 days (3× improvement) by developing CLI tools and automated workflows for test set creation, model inference, data labeling, and performance metric computation.
• Designed and implemented an automated monitoring workflow to track predictive performance of multimodal models in production, integrating data sampling, human-in-the-loop labeling (SageMaker Ground Truth), performance evaluation, and generation of HTML audit reports with minimal human intervention.
• Designed, executed, and analyzed experiments to evaluate multimodal models (e.g. CLIP) for text-to-image retrieval use cases, including search recommendations and customer review media.
• Designed and implemented an event-based system to filter Amazon product updates for book-specific changes in near real-time, now utilized by multiple teams and processing over 6K updates per second.
• Led the productization of a textbook binary classification model, including the design and implementation of pipelines for training, evaluation, deployment, and monitoring. This model, which is integrated into Amazon's ingestion pipeline, classifies over 4M book updates daily.
• Implemented a prototype of a RAG / Q&A bot for Kindle Books using OpenAI embeddings and LLMs. This prototype won the Madrid Books 2022 hackathon, among 15+ competing projects.
• # Key Technologies: AWS Stack (SageMaker, Lambda, SQS, S3, DynamoDB, CDK, CloudWatch, Step Functions, Ground Truth...), Java, Python, Docker, Spark Scala, AutoGluon

Sep 2018 – Sep 2020 · 2 yrs · Madrid Area, Spain

• I worked with clients in the banking sector and had the following main responsibilities:
• Contributed to the development of an MLOps platform for streamlining the training, evaluation, and deployment of machine learning models. This platform was turned into a project accelerator that Teradata now uses in multiple international client engagements.
• # Key Technologies: Java, Python, Spring Boot, ActiveMQ, Gitlab CI/CD, Docker, Kubernetes, S3, MySQL
• # More info: https://tinyurl.com/t98xcnf2
• Led a team of 3 data engineers and collaborated with data science, DevOps, and UI/UX teams in a consulting project. The goal was to build a system that could automatically classify and extract data from documents using OCR and NLP techniques. My responsibilities spanned the full software development life cycle, including task planning, coding and code review, API design, testing, monitoring, and documentation writing. Currently, this system processes over 14K documents per day.
• #Key Technologies: Java, Spring Boot, REST, ActiveMQ, Docker, Kubernetes, S3, Teradata DB, TesseractOCR
• Led the operationalization of an ML model for detecting fraudulent bank operations with a 1-second SLA, as part of a consulting project. My responsibilities included overseeing the full software development life cycle, including project planning, code development and review, API design, testing, monitoring, and documentation. This system currently analyzes up to 10 million operations per day.
• #Key Technologies: Java, Spring Boot, REST, Docker, Kubernetes, JMeter
• Set up a datalab environment for enabling data scientists to easily explore large datasets and prototype machine learning models.
• #Key Technologies: Ansible, JupyterHub, Anaconda, Python

Jul 2016 – Jun 2018 · 1 yr 11 mos · Braga Area, Portugal

• Member of the HASLab group. Conducted research on different aspects of distributed systems, namely monitoring, debugging, and large-scale storage. Research work was published in several international conferences: DAIS’18, DSN’18, AAAI’19, OPODIS'19, SRDS’19, ICDCS'20.
• #Google Scholar: https://scholar.google.com/citations?user=mitxUPAAAAAJ&hl=en

May 2015 – Jun 2015 · 1 mo

• Designed and implemented Cortex, a tool that automatically discovers latent concurrency bugs in Java multithreaded applications through symbolic execution and SMT constraint solving.
• #Website: https://github.com/nunomachado/cortex-tool
• #Paper: "Production-guided Concurrency Debugging". Published in PPoPP '16.

May 2014 – Aug 2014 · 3 mos · Redmond

• Designed and implemented Symbiosis, a tool that automatically isolates concurrency bugs in Java and C/C++ multithreaded applications through symbolic execution and SMT constraint solving.
• #Technologies: Java, C/C++, Soot, LLVM, Java Path Finder, KLEE, Z3
• #Website: https://github.com/nunomachado/symbiosis
• #Paper: "Concurrency Debugging with Differential Schedule Projections". Published in PLDI '15.

Sep 2010 – Jan 2011 · 4 mos

• Contributed to the development of the fault replication component of the FastFix platform (FP7 European Project). This component is responsible for monitoring an application at the client side and generate lightweight execution logs that allow reproducing errors at the maintenance side.
• #Technologies: Java, OSGI, Soot
• #Website: http://fastfixrsm.sourceforge.net/