ML Engineer - Verifications

Kodex•,

1d•$150,000 - $180,000•Remote

About The Position

Kodex is seeking an ML Engineer to join their Verifications / Threat Intel team. This role will focus on building the intelligence layer that protects the platform by detecting suspicious activity, improving verification accuracy, and creating scalable, auditable systems from real-world data. The position is at the intersection of product, data, and security, requiring close collaboration with Threat Intel operators and engineers to translate investigative workflows into production-grade models, pipelines, and decision-support tooling. The goal is to move beyond simple heuristics towards systems that are measurable, explainable, and safe to operate.

Requirements

4+ years of experience building software ML systems in production.
Experience in fraud, abuse, trust & safety, risk, or security analytics.
Experience with modern ML/DS workflows: experimentation, evaluation, and deploying models that other teams rely on.
Comfortable with data pipelines and messy real-world data, including instrumentation gaps, bias, label noise, and changing definitions.
Careful consideration for reliability and safety, including rollbacks, guardrails, human-in-the-loop review, and auditability.
Clear communication across disciplines and ability to translate ambiguous investigative needs into well-scoped engineering work.
Pragmatic approach, knowing when to start with a baseline and when to invest in stronger modeling/infrastructure.

Nice To Haves

Familiarity with LLMs / RAG / agentic workflows for internal tooling or decision support.
Experience in compliance-sensitive environments, including privacy, access controls, and audit trails.

Responsibilities

Work across the full ML lifecycle: problem framing, data exploration, deployment, monitoring, and iteration.
Build and deploy models for detecting and flagging suspicious behavior using classification, anomaly detection, clustering, ranking, and other appropriate methods.
Own ML pipelines end-to-end, including feature generation, training, evaluation, batch/streaming inference, backfills, and versioning.
Design evaluation and monitoring strategies, including ground truth definition with Threat Intel, offline metrics, drift monitoring, and alerting.
Collaborate with product and engineering teams to integrate intelligence into user-facing workflows like review queues and decision-support tooling.
Improve data quality and accessibility by applying data engineering patterns such as schemas, lineage, reproducibility, and access controls.
Contribute to operational rigor through runbooks, incident response, and safe rollout practices for systems impacting customer trust.