avra is a venture capital firm that works with growth-stage founders as they scale beyond Series A. The firm provides strategic investment and practical support to help portfolio companies develop their businesses, with a focus on backing innovation and entrepreneurship across the B2B and finance landscape.
Head of AI Research — Remote in Brazil
Lead AI research for Avra’s Brazilian enterprise decision platform, focusing on graph-based foundation models. Guide scientists and engineers while turning advanced research into reliable production systems.
Description
- Set the foundation-model research strategy across knowledge graphs, representation learning, and self-supervised or unsupervised methods
- Recruit, mentor, and develop a team of research scientists and engineers
- Create a disciplined research process spanning hypotheses, RFCs, experiments, and evidence-based decisions
- Develop advanced graph neural networks for large-scale graph applications
- Address node-, edge-, and graph-level problems, including multi-scale embeddings and temporal or inductive generalization
- Build dependable, reproducible training and evaluation workflows with PyTorch, PyG, and distributed-training frameworks
- Create and maintain rigorous benchmark suites for statistically valid model comparisons
- Work with Product Engineering to deliver models for batch and real-time inference
- Collaborate with go-to-market teams on enterprise success measures and explain business outcomes to technical and executive audiences
- Manage concept drift and strengthen deployment approaches for regulated environments
Requirements
- At least seven years of AI/ML experience, or a PhD combined with four years of relevant experience
- Track record of moving research from papers or prototypes into scalable, dependable software with measurable business results
- Advanced hands-on Python and PyTorch capability
- Strong command of graph-learning libraries such as PyTorch Geometric or DGL
- Sound software engineering practice, including testing, profiling, and maintainable system design
- Experience mentoring technical teams, leading major projects, or managing a team of two to six scientists or engineers
- Professional fluency in Portuguese and English
- Distributed training and inference experience is preferred
- Experience developing graph-based models is preferred
- Knowledge of graph-focused self-supervised or contrastive learning, plus either leading AI publications or substantial open-source work, is preferred
- A relevant degree or PhD is required