Senior ML Engineer

Qode•Ontario, ON

5d•Onsite

About The Position

We are looking for a Senior ML Engineer to design, build, and productionize ML pipelines for a Trust Scoring platform, with a strong focus on replayability, determinism, explainability, and MLOps best practices. This role is hands‑on and platform‑focused, working across batch inference, real‑time scoring, feature engineering, and model monitoring, within an AWS‑native architecture.

Requirements

3–5 years hands‑on experience as a Machine Learning Engineer
Strong experience taking ML models from development to production
Programming: Python, PySpark
ML/MLOps: MLflow, Model versioning and promotion, Drift detection and monitoring
Data: Feature engineering, Batch and streaming concepts, Large‑scale datasets
AWS experience (preferred): S3, Spark/EMR, IAM, basic networking
Familiarity with feature stores
Familiarity with API‑based inference patterns

Nice To Haves

Experience with fraud, trust scoring, or risk modeling
Exposure to PII‑sensitive systems
Experience migrating batch ML pipelines to real‑time scoring
Knowledge of explainable ML techniques

Responsibilities

Productionize PoC ML models into reproducible, governed pipelines
Implement deterministic preprocessing for train vs serve parity
Develop batch and near‑real‑time inference workflows
Generate explainability artifacts (reason codes, score attribution)
Implement and maintain MLflow (experiments, model registry)
Implement and maintain CI/CD pipelines for ML
Implement and maintain Champion/Challenger model frameworks
Enable controlled rollouts (shadow, advisory, active scoring)
Enable versioned feature and model deployments
Design and consume features from batch and low‑latency feature stores
Design and consume features from canonical entity models (subscriber, device, SIM)
Collaborate on data quality validation
Collaborate on schema contracts
Collaborate on drift detection (feature + score)
Implement feature drift detection
Implement model performance monitoring
Implement SLA and freshness validation
Support replay and recovery using idempotent design patterns