Machine Learning Engineer

The role

You will design, build, and evaluate language models that power AI security products. You will own the end-to-end pipeline from dataset curation to model integration into production systems.

What you will do

• Build scalable pipelines to collect, preprocess, and manage datasets for training and evaluation of LLMs.
• Design and run experiments to evaluate LLMs on accuracy, robustness, fairness, and safety.
• Create dashboards, reports, and visualizations to communicate evaluation results, trends, and failure cases.
• Develop and leverage knowledge graphs to structure data and improve model performance.
• Work with researchers to translate new ideas into engineering workflows.
• Work with data scientists to automate QA checks and guardrails.
• Fine-tune, optimize, and integrate models into production systems.
• Contribute to ML tooling and experimentation frameworks.

Required qualifications

• 2–5+ years working in machine learning or data science
• Strong software engineering background (Python, testing frameworks like pytest/unittest, CI/CD tools)
• Proficiency in ML frameworks such as PyTorch
• Experience with data engineering tools (e.g., Spark, Kafka, Airflow)
• Familiarity with deploying models on cloud platforms (AWS, GCP, or Azure) and containerized environments (Docker, Kubernetes)
• Strong knowledge of ML fundamentals (supervised/unsupervised learning, deep learning, NLP)
• Interest or background in cybersecurity, adversarial ML, anomaly detection, or related fields
• Comfortable working in fast-moving, ambiguous environments

Nice to have

• Research or industry experience in adversarial ML, model robustness, or explainable AI
• Experience building interactive dashboards for model monitoring and visualization
• Contributions to open-source ML, NLP, or security projects

About the company

Witness AI invented intent-based AI security. They provide visibility into how employees and systems use AI to help security teams monitor risk and enforce guardrails.