Overview
Intuit is looking for a Staff Machine Learning Engineer to own the data and platform layer beneath our consumer risk decisioning. This team builds the machine learning that decides, in real time, whether money moves — protecting customers from account takeover, first/third-party fraud, and unauthorized transactions across Intuit Fintech products.
You'll design and own the shared infrastructure every model on this team depends on: streaming and batch feature pipelines, the cross-entity data path, training and evaluation frameworks, real-time inference serving, and the handoff into our decision engine, across entire Fintech money product lifecycle — from risk screening/qualification of money-in/out events, cashflow underwriting, account take-over detection, to dynamic segmentation. What you architect becomes the reference pattern the whole organization adopts.
Responsibilities
- Own the technical vision and architecture for the consumer risk data and serving platform — feature pipelines, the cross-entity data path, training and evaluation infrastructure, and real-time inference — balancing tradeoffs and long-term implications
- Design and build multi-cloud infrastructure stack and enable data handshake, working through federated account link mapping, and land curated, governed datasets in the Intuit's central data lake
- Build shared feature infrastructure spanning streaming and batch, consumed by multiple model workstreams, with the observability to catch drift and staleness
- Establish evaluation frameworks that make model quality, regression, and production impact measurable across the team's portfolio
- Own the model-to-decision path: deployment through model serving, integration with the decision engine, and correctness and latency on a sub-second decision budget
- Set and enforce engineering standards for ML systems on this team — testing, observability, reproducibility, operational excellence — and structure codebases for agent-assisted development and autonomous navigation
- Engineer closed-loop workflows that automate the repetitive parts of the model lifecycle, moving beyond point automation toward orchestrated systems needing minimal intervention
- Eliminate barriers caused by technical and prioritization complexity, including dependencies that cross into platform and data teams — cultivate the partnerships that make those dependencies tractable
- Generalize what you build into a reference pattern other teams can adopt rather than rebuild, and document it so the pattern travels
- Mentor engineers on ML systems craft; provide actionable feedback to senior engineers and help them break work into pieces agents can execute reliably