SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: CAD 240,000 - 270,000 / annual
Narvar is seeking a Director of Machine Learning to lead the ML organization behind post-purchase commerce systems serving hundreds of millions of consumers across 20B+ orders and 1,300+ retail brands. You will own the machine learning charter across identity resolution (Graphite engine), fraud and abuse detection (IRIS engine), risk scoring, and consumer intelligence—including strategy, roadmap, and delivery.
Day-to-day responsibilities include:
- Building and growing a high-performing, globally distributed team of ML engineers; hiring, coaching, and developing senior individual contributors and managers
- Setting technical standards for model development, evaluation, deployment, and monitoring
- Driving identity resolution coverage, precision, and profile classification accuracy against measured baselines
- Owning IRIS model performance end-to-end: detection rate, false-positive rate, label quality, and feedback loops as fraud patterns evolve
- Building the ML platform layer (feature stores, training pipelines, model registry, online serving, drift and performance monitoring) to prevent infrastructure bottlenecks
- Partnering with AI engineering to ensure identity and risk signals are first-class inputs to NAVI agent decisions
- Translating retailer problems into ML problems worth solving; collaborating with Product, Engineering, Security, and Customer Success
- Owning build-vs-buy and data-partner decisions, including economics
- Communicating model performance, risk, and tradeoffs to executives, retailers, and the board
Why this role matters: Narvar's cross-retailer visibility across 20B+ orders creates a unique dataset—coordinated fraud patterns invisible to single retailers become visible here. The problem is adversarial and unsolved; returns fraud grows faster than retail itself, and adversaries adapt quarterly. Stakes are measurable (fraud loss, false-positive cost, identity coverage, resolution rate). As AI shopping agents enter commerce, identity and risk form the trust layer for returns, claims, and delivery promises. Much of this work is greenfield: real-time resolution, cold-start modeling, historical backfill, and new data partnerships are decisions still being made.
Requirements:
- 12+ years in engineering with 5+ years managing ML or data teams, including managing managers or senior individual contributors
- Strong engineering foundation: can read training pipelines, review feature specs, and identify flawed evaluations
- Shipped ML systems making consequential automated decisions in production; owned them post-launch (drift, retraining, incidents)
- Deep experience with at least one of: entity resolution / identity graphs, fraud and abuse detection, risk scoring, or anomaly detection at scale
- Understanding of ML-specific challenges: noisy/delayed labels, silent failures, offline metrics limitations, adversarial adaptation
- Strong opinions on evaluation: precision/recall tradeoffs on imbalanced data, holdout hygiene, feedback loop contamination
- Built or scaled ML infrastructure: feature pipelines, training orchestration, online serving with latency budgets, monitoring and alerting
- Fluent in Python; comfortable with modern data stack (Spark, Airflow or equivalent, streaming, cloud data warehouses; company uses GCP)
- Proven hiring and retention track record; can point to engineers whose careers improved under your leadership
- Thrive in fast-paced, flat-structure environments; set direction through earning trust, not mandate
- BS/MS in computer science, statistics, or equivalent background
Bonus experience:
- Identity resolution or graph systems against third-party consumer data (credit bureau, telecom, etc.), including matching logic and confidence scoring
- Adversarial ML systems where attackers actively adapt to detection
- Real-time inference paths with identity/risk resolved within request budgets
- Retail, payments, fintech, insurance, or marketplace trust & safety background
- Close partnership with LLM/agent teams; understanding how structured ML signals ground agentic decisions
- Privacy, consent, and data-governance constraints on consumer identity data
- 0→1 experience: standing up functions, platforms, or data partnerships from scratch
- Modern tooling including AI-assisted development workflows to increase leverage
