Applied Data Scientist

Varonis

29d

About The Position

Varonis is seeking a highly skilled and motivated Applied Data Scientist to design, build, evaluate, and deploy machine learning, LLM, and agentic capabilities that improve cybersecurity detection, investigation, and response workflows. This hands-on applied role uses data science, machine learning, LLMs, agents, and modern AI systems to address practical cybersecurity problems. The Applied Data Scientist will work closely with data and software engineers, security researchers, and threat analysts to turn complex security data and attack scenarios into reliable, scalable, measurable AI-powered product capabilities. The position includes developing and evaluating AI workflows, partnering on production deployment, monitoring performance, and improving capabilities based on operational results and user needs.

Requirements

Bachelor’s degree in Computer Science, Engineering, Data Science, Statistics, or a related field.
Strong hands-on foundation in machine learning, applied statistics, and data science, including supervised learning, unsupervised learning, anomaly detection, model evaluation, and experimentation.
At least 3 years of experience applying data science or machine learning in production or applied settings.
Proficiency in Python and common data science and machine learning libraries.
Experience working with LLMs in applied or production contexts, including prompt engineering, model selection, evaluation, retrieval-augmented generation, and safe deployment.
Familiarity with embeddings, vector databases, retrieval systems, and RAG-based workflows for security, knowledge-intensive, or analyst-facing applications.
Understanding of AI and LLM security considerations, including adversarial inputs, prompt injection, data privacy, model misuse, governance, and safe system design.
Experience partnering with engineering teams to deploy, monitor, and improve ML models, AI workflows, or data products in production environments.
Ability to reason under uncertainty, work with noisy and incomplete data, and make pragmatic tradeoffs among model performance, explainability, latency, reliability, and operational value.
Demonstrated ability to independently frame ambiguous problems, develop and validate solutions, and drive them from experimentation through production and measurable impact.
Strong communication and collaboration skills, including the ability to work effectively across security, engineering, data, product, and research teams.

Nice To Haves

A master’s degree is a plus.
Experience in cybersecurity, fraud, risk, abuse, observability, or other adversarial or high-noise domains is strongly preferred.
Experience with PySpark, Databricks, SQL, or large-scale data processing frameworks.

Responsibilities

Apply machine learning, statistical analysis, LLMs, and agent-assisted techniques to cybersecurity problems in detection, triage, investigation, and response.
Analyze large and complex security datasets to identify behavioral patterns, anomalies, attack signals, and trends that can improve threat detection and analyst workflows.
Translate cybersecurity use cases into data science problems, including problem framing, dataset creation, feature development, model selection, evaluation, and iteration.
Design, develop, and evaluate ML- and LLM-powered security workflows, including retrieval, reasoning, tool use, human-in-the-loop review, feedback loops, and guardrails.
Build evaluation frameworks, metrics, benchmarks, and test datasets to measure model quality, reliability, precision, recall, latency, robustness, and operational impact.
Develop prompts, instructions, retrieval strategies, and model interaction patterns to improve the usefulness, consistency, and safety of LLM-powered features.
Partner with software engineers to productionize models, AI agents, and data pipelines in secure, scalable, and maintainable systems.
Monitor deployed models and workflows for performance drift, data quality issues, false positives, false negatives, and opportunities for continuous improvement.
Collaborate with cybersecurity researchers, threat analysts, and product stakeholders to ensure AI capabilities address user needs and evolving threat scenarios.
Translate relevant advances in machine learning, LLMs, agentic AI, and cybersecurity into practical product improvements, evaluation methods, and internal best practices.