Titulo Estágio
Design and Evaluation of Resource-Efficient Machine Learning Systems for Real-Time Fraud Detection
Local do Estágio
Remote, Hybrid or Lisbon/Porto/Coimbra office
Enquadramento
Feedzai provides cutting-edge transaction monitoring solutions powered by advanced machine learning. As part of our innovation roadmap, we're designing a next-generation fraud detection system that is lean, cost-efficient, and capable of integrating modern AI workflows such as deep learning and generative models. This project contributes to the ongoing effort within our AI Research team to rethink the architectural and algorithmic foundations of our platform. It explores efficient systems for stateful event processing, incremental learning, and approximate computations that allow us to maintain performance while reducing compute and memory footprint.
The intern will work under the supervision of senior AI researchers and collaborate with a multi-disciplinary team to explore, prototype, and evaluate novel techniques, like quantised feature representations, probabilistic data structures, and event-driven Bayesian modeling, that power resource-conscious fraud detection pipelines. This int
Objetivo
The primary objective of this internship is to contribute to the development and evaluation of a new generation of fraud detection systems that are both performant and resource-efficient. Interns will engage in applied research and prototyping tasks that directly inform the product roadmap of Feedzai’s AI tools and infrastructure. Key learning and performance goals include:
- Prototype Development: Design and implement experimental fraud detection modules using novel ML/DL techniques that are compatible with lean data aggregation and storage architectures.
- Research and Innovation: Explore concepts including quantised representations, probabilistic data structures, stateful learning, and approximate computing, with the goal of identifying techniques that balance cost and accuracy.
- Scientific Contributions: Participate in the preparation of technical white papers, conference papers, or internal technical documentation. Depending on the maturity of the work, interns may co-author scientific publications or invention disclosures.
- Evaluation and Benchmarking: Carry out meaningful evaluations of performance, efficiency, and cost trade-offs for different ML workflows using real or synthetic data representative of high-volume transaction monitoring environments.
- Collaborative Research Practice: Gain experience working within a cross-functional research team composed of ML researchers, data engineers, and systems architects.
Plano de Trabalhos - Semestre 1
The first semester will be exploratory and designed to accommodate the student’s academic commitments. Activities will focus on background study, light experimentation, and early-stage contributions that lay the foundation for more intensive work in the second semester.
- Contextual Immersion: Understand the scope of resource-efficient machine learning in the context of fraud detection. Explore internal documentation on current systems and research directions.
- Literature Review: Conduct a focused review on selected topics, with examples including: quantised freature representations, probabilistic data structures, approximate computing, and stateful event processing.
- Prepare a short research memo summarising relevant academic and industry approaches.
- Exploratory Experiments: Design small-scale experiments and document initial insights and ideas for scalable application.
- Proposal for Second Semester Work: Define a research question or hypothesis for deeper exploration.
This phase is intended to encourage critical thinking, creativity, and familiarity with both the problem domain and Feedzai’s research environment, while being mindful of the student’s availability.
Plano de Trabalhos - Semestre 2
The second semester will be dedicated to hands-on research and development under full-time engagement. The student will work closely with the AI Research team to prototype, benchmark, and document novel approaches for resource-efficient ML systems.
- Prototype Design and Implementation: Develop end-to-end prototypes that showcase efficient ML techniques, such as Hashed or quantised feature representations, Event stream filtering and state
aggregation, Incremental model retraining workflows, ... Using representative fraud datasets to test effectiveness.
- Evaluation and Benchmarking: Support the design of evaluation pipelines to compare performance across key dimensions: accuracy, latency, memory usage, computing costs.
- Contributions to IP and Scientific Outputs: Support the drafting of internal technical reports, invention disclosures, or scientific papers. Where appropriate, contribute as co-author or inventor to publications or patents.
This phase emphasizes real impact, measurable contributions, and research excellence, while offering the intern the opportunity to work on a real-world problem with deep technical and societal relevance.
Condições
Interns may work remotely or from Feedzai’s offices in Lisbon, Porto or Coimbra (hybrid models supported). We offer flexible working hours.
The intern will receive dedicated mentorship by senior AI researchers, with regular one-on-one check-ins and weekly team syncs. Access to internal ML platforms, compute infrastructure, and proprietary datasets will be offered.
All scientific contributions will be acknowledged in patents and publications where applicable
Observações
The internship offers exceptional exposure to real-world problems and constraints in the FinTech domain, with a strong emphasis on AI for good and responsible AI practices.
Candidates with interests in ML research, efficient computing, and systems design are especially encouraged to apply. Past interns in this team have transitioned into research scientist roles or pursued PhDs with joint supervision from Feedzai staff.
Orientador
Iker Perez, Jacopo Bono
iker.perez@feedzai.com, jacopo.bono@feedzai.c 📩