Head of Research (AI)

ABOUT AVRA
Avra is building relational foundation models for enterprise decision-making in Brazil.
Our work focuses on graph-native models for structured, high-stakes prediction problems: credit, fraud, growth, monitoring, and other decisions where entities cannot be understood in isolation. We model companies, people, and the relationships between them as evolving networks, then adapt those representations to customer-specific prediction tasks that plug into existing decisioning systems.
We work with internationally recognized research advisors, and we care about research that becomes useful in production.
THE ROLE
This is a leadership role for a hands-on "player-coach" who will own and elevate our research function. Your mission is to scale the scientific vision for our Foundation Models for Relational Data (e.g., graphs, databases). You will raise the technical bar for the entire team, and bridge the gap between cutting-edge research and production-grade capabilities that move our clients' most important metrics. You will partner deeply with our Data teams (graph & data pipelines), Product Engineering & MLOps to transform ambitious ideas into the core of Avra's platform.
WHAT YOU'LL DO
LEADERSHIP & SCIENTIFIC STRATEGY
- Own the research roadmap for our Foundation Models, focusing on knowledge graphs, representation learning, self-supervised and unsupervised methods.
- Lead, mentor, and grow a world-class team of research scientists and engineers.
- Establish and champion a rigorous research cadence: from hypothesis definition and RFCs to disciplined experimentation and clear, data-driven decision-making.
HANDS-ON RESEARCH & ENGINEERING
- Design and implement state-of-the-art GNNs for our unique, large-scale graph. Solve complex problems in node, edge, and graph-level tasks, multi-scale embeddings, and temporal/inductive generalization.
- Build reliable, reproducible training and evaluation pipelines using PyTorch, PyG, and distributed training frameworks.
- Define and maintain our gold-standard benchmarks, ensuring statistically sound model comparisons.
PRODUCTION & DELIVERY
- Collaborate with Product Engineering to productize models to serve both batch and online inference to our customers.
- Partner with our GTM teams to define success criteria for enterprise clients, and to communicate the impact of your team's work to technical and executive stakeholders.
- Ensure our research accounts for the challenges of real-world systems, including concept drift, and improve our strategy for model deployment in regulated contexts.
YOU SHOULD HAVE
- 7+ years of experience in AI/ML (or a PhD + 4 years)
- A demonstrated ability to ship research into production: you have taken ideas from a paper or prototype to scalable, reliable code that delivered measurable business impact.
- Hands-on excellence in Python and PyTorch, with deep proficiency in graph learning libraries like Pytorch Geometric or DGL.
- Solid software engineering fundamentals, including testing, profiling, and building maintainable systems.
- Proven experience mentoring and leading technical projects or managing a small team (2-6) of scientists/engineers.
- Professional proficiency in both Portuguese and English.
PREFERRED QUALIFICATIONS
- Experience with distributed training and inference.
- Experience working with graph-based models.
- Deep knowledge of self-supervised or contrastive learning techniques, particularly for graphs.
- A strong publication record in top-tier AI conferences (NeurIPS, ICML, ICLR, etc.) or significant open-source contributions.

Employment type
FullTime

Work arrangement
Yes