Lead Machine Learning Engineer - LMTS

Salesforce•San Francisco, CA

3d•Hybrid

About The Position

Salesforce is seeking a Senior / Lead member of technical staff for Machine Learning Engineering. This role is part of the foundation machine learning platform team within the Trust Intelligence Platform organization, focusing on building and accelerating scalable and resilient machine learning pipelines across the security engineering organization. The ideal candidate is a highly motivated, hands-on lead machine learning engineer with a strong business understanding of cybersecurity problems, acting as a force multiplier security data scientist for the security organization. This role involves architecting the data-driven strategy for threat detection capabilities, translating security threats into mathematical problems, and championing a rapid prototyping culture to validate hypotheses quickly. The work will focus on evolving threat detection using advanced probabilistic modeling, graph analytics, and supervised/unsupervised learning to expose sophisticated threats that evade traditional defenses. Additionally, the role involves mentoring junior scientists and engineers, building internal tooling, feature stores, and libraries, and influencing the broader security engineering roadmap to ensure closed-loop security telemetry. The position requires delivering production-grade models with engineering rigor (CI/CD, scalable code) and adversarial resilience to minimize alert fatigue and maximize analyst efficiency.

Requirements

3-5+ years in data science, with at least 2+ years dedicated to the cybersecurity domain designing, implementing, and deploying systems of anomaly detection, clustering, and graph models in production.
Hands-on comfort with high-volume logs.
Proficiency with Spark/Pyspark, Snowflake, Flink, and streaming services such as Apache Kafka.
Deep understanding and application of containerization (Docker) and workflow orchestration (Kubernetes, Apache Airflow) for automated ML pipelines.
Mastery of Python programming, including proficiency in leading ML frameworks (TensorFlow, PyTorch).
Adherence to software engineering best practices.
Demonstrated success in implementing comprehensive MLOps methodologies, encompassing CI/CD pipelines, testing protocols, and model performance monitoring.
Solid foundation in feature engineering techniques and the implementation of feature stores.
Experience in formulating ML governance policies and ensuring adherence to data security regulations.
Ability to explain complex statistical concepts to non-technical stakeholders and executive leadership.
Proven ability to manage scope, timelines, and stakeholder expectations across multiple organizations.
High degree of autonomy with the ability to look at a vague business problem and structure a data-driven solution without needing a predefined roadmap.

Nice To Haves

Masters or PhD in a quantitative field.
Expertise in advanced Natural Language Processing (NLP) methodologies.
Experience contributing to open-source security data science tools.
Presentations at major security conferences (Black Hat, DEF CON, BSides) or data conferences.
Background in offensive security (Penetration Testing/Red Teaming) with an "attacker's mindset."
Demonstrated experience conducting research or working collaboratively with Machine Learning (ML) research teams.
Previous experience in a mentoring role for junior engineers.
Track record of publications and/or patents in quantitative disciplines.

Responsibilities

Own the decision-making process for translating vague security threats into concrete mathematical problems.
Champion a rapid prototyping culture to validate hypotheses in days rather than months.
Lead the evolution of threat detection by introducing advanced probabilistic modeling, graph analytics, supervised and unsupervised learning.
Expose sophisticated threats such as active system intrusions, lateral movement, beaconing, and insider attacks.
Act as a force multiplier by mentoring junior scientists and engineers.
Build internal tooling, feature stores, and libraries to increase team speed.
Influence the broader security engineering roadmap to ensure closed-loop security telemetry is treated as a first-class citizen.
Deliver production-grade models with engineering rigor (CI/CD, scalable code) and adversarial resilience.
Minimize "alert fatigue" and maximize analyst efficiency.