3 hours ago
Base Salary
$220k - $260k/yr
Responsibilities
- Turn messy customer data into production models and choose appropriate classical, LLM-based, or third-party approaches.
- Improve classification pipelines, confidence cascading, and harmful-content detection strategies while balancing cost, latency, and accuracy.
- Develop intelligent features that assist moderators, organize platform content, and reveal data patterns.
- Build Cinder’s model training, hosting, and inference platform with Engineering.
- Design evaluation and metrics infrastructure for classifier scores and model outputs.
- Shape evaluation architecture for AI agents and measure decision quality, tool usage, and cost.
- Partner with the Data Engineer on data infrastructure, training data, feature pipelines, and production inference.
- Mentor teammates and raise the company’s machine learning standards.
Requirements
- 5–8+ years of machine learning engineering experience on a small team.
- Demonstrated experience shipping gradient-boosting, tree-based, classifier, or embedding-based ML systems to production.
- Experience taking classification problems from messy, unlabeled real-world data through deployment and ongoing production use.
- Ability to evaluate when LLMs are worthwhile compared with classical models based on cost and latency.
- Hands-on experience building classifiers under severe class imbalance.
- Experience creating training pipelines, serving infrastructure, evaluation harnesses, and monitoring from scratch.
- Startup or small/mid-size company experience with meaningful ownership and pragmatic build-versus-buy decisions.
- Strong knowledge of feature engineering, leak-aware train/test splits, imbalanced-data metrics, cross-validation, and hyperparameter tuning.
- Strong Python skills and hands-on experience with PyTorch, scikit-learn, LangChain, and XGBoost.
- MLOps experience including ML CI/CD, model versioning, experiment tracking, drift detection, and production monitoring.
- Experience designing inference systems with explicit latency and throughput targets and balancing model complexity, cost, and performance.
- Databricks experience is a bonus.
- AWS and Terraform experience is a plus.
Benefits
- NYC-based role requiring relocation and in-person collaboration.
- At least two all-company events per year.
- Health, vision, and dental benefits.
- 401(k) plan with employer matching.
- Fully paid commuter benefits.
- Fully stocked office with paid lunch and dinner.
Categories
About Cinder
Cinder is the industry’s first Trust and Safety operations platform to help organizations combat Internet abuse at scale. We provide Trust and Safety teams with a single system to manage complex integrity operations and investigations to create a safe environment for their users.