AI Research Manager - Machine Learning

Nubank•Palo Alto, CA

3d•Hybrid

About The Position

As AI Research Manager, you will own the operating health and execution of Nubank's ML research agenda. This is a dedicated people-management role for a small, high-talent-density team of world-class researchers, partnering closely with senior technical leads who drive the scientific direction. The team works on a portfolio of research bets on a one-to-four-quarter horizon. This work compounds into the production capabilities Nubank will depend on. Current and upcoming bets include: Next-generation nuFormer architectures: proprietary transformers that learn from raw transaction sequences and power key production credit decisions. Multitask and multi-target modeling: single models serving many high-impact prediction tasks at once. Training and inference efficiency: distillation, quantization, sparsity, and parallelism to run state-of-the-art models economically at our scale. Causal modeling and policy optimization: moving beyond prediction to the decisions and policies those predictions should drive. World models: open-ended models that reason about a customer's full financial life. Recommendation systems: extending the backbone to app events, engagement, and personalization signals. Embeddings and representation learning: semantic IDs, contrastive learning, and reusable representations used across the bank. Real-time and continual learning: low-latency inference and models that adapt over time. Your job is to make exceptional science happen: build the conditions, focus, and operating rhythm that researchers do great work.

Requirements

3+ years of direct people-management experience leading applied AI research or core machine learning teams at big tech companies, frontier labs, or comparable high-scale environments.
M.S. or Ph.D. in Computer Science, Machine Learning, Applied Mathematics, or a related quantitative field, with a strong working understanding of core AI methodologies.
Proven ability to build, scale, and retain high-performing research or advanced applied-science teams.
Deep conceptual familiarity with at least one of our core research areas: Foundation models & LLMs: pre-training from scratch, scaling laws, training-optimization frameworks, and large GPU-cluster workflows.
Behavioral & sequential models: sequence models, recommendation systems, and large-scale representation learning for very large user bases.
Decisioning & optimization: causal inference, policy optimization, constrained optimization, or reinforcement learning.
Training & inference efficiency: model sparsification, quantization, distillation, or parallelism and partitioning design.
Exceptional storytelling skills, with a track record of translating complex technical milestones into business impact for senior and C-level audiences.
Experience operating in global, distributed teams and navigating the path from foundational research to scalable, production-grade systems.

Responsibilities

Lead, mentor, and advocate for a team of world-class ML researchers, fostering an environment of psychological safety, high ambition, and rigorous scientific inquiry.
Own the operating health of the team, including performance, career growth, hiring, and compensation cycles for elite individual contributors who often operate at staff-and-above technical depth.
Attract and retain top-tier research talent in a competitive market, and build a reputation for the team as a place the best researchers want to be.
Own the operating cadence of a portfolio of two-to-three concurrent, quarter-scale research bets, from problem framing and OKRs through progress tracking and clear go/no-go decisions.
Allocate scarce, high-value resources, most notably GPU capacity, across competing research priorities, balancing exploration against the bets most likely to compound.
Protect deep-focus research time. Sustaining a long-term agenda in a fast-moving company means deliberately creating the space for rigorous, multi-quarter work, so the team can pursue ambitious bets instead of being fully absorbed by short-term applied demands.
Raise the bar on research rigor and communication: strong experimental design, peer review, and reproducibility.
Cultivate the team's standing in the broader research community. Real-world impact is our primary measure of success, but we actively encourage publishing, open-source contribution, and conference presence that build a reputation reflecting the quality of the work and help attract the best researchers.
Partner closely with senior technical leads who drive architecture and scientific direction, aligning operational execution with the long-term research roadmap so that scientific and operating decisions reinforce each other.
Ensure clean handoffs from research into production, so breakthrough results land reproducibly in the hands of applied teams rather than stalling as one-off experiments.
Translate complex, frontier technical progress (training efficiency, causal inference, novel architectures) into clear, high-impact narratives for executive leadership, connecting research milestones to business outcomes.
Bridge foundational research and long-term business strategy, ensuring the team's technical inputs directly enable Nubank's AI-first direction.