Applied ML Engineer

Applied ML Engineer – Backend & Agentic Security

Full-Time   ·   San Francisco, CA  

About CompFly AI

CompFly is building the security layer for the autonomous AI era; real-time defense for AI agents that act, decide, and operate on behalf of humans inside enterprise environments. As autonomous agents become the new compute primitive, CompFly ensures enterprises can trust what those agents do. We are a small team that moves fast and builds things that matter.

The Role

We are looking for an Applied ML Engineer who thrives at the intersection of research and production engineering. You will design and train ML models for agentic threat detection, build the backend infrastructure that serves them at low latency, and ship systems that operate at the speed of autonomous agents. This is not a pure research role or a pure backend role. You will do both. The ideal candidate is someone who can go from training a detection model one week to building the inference pipeline that deploys it the next.

What You'll Do

• Design, train, and evaluate ML models for agentic threat detection (adversarial prompt classification, anomalous tool call patterns, behavioral drift detection, and cross-agent coordination analysis).
• Build and own the ML backend infrastructure: model serving, real-time inference pipelines, feature stores, and scoring APIs written in Rust and Go.
• Develop novel detection methods for autonomous agent behaviors including multi-turn interaction analysis, privilege escalation patterns, and data exfiltration signals.
• Build feedback loops that capture production signals and continuously improve model accuracy and recall.
• Contribute to dataset construction (synthetic data generation, labeling pipelines, and benchmark design for agent security scenarios).
• Evaluate frontier LLM behaviors and contribute to internal adversarial assessments of model-layer risks.

What We're Looking For

• 3+ years of applied ML engineering with demonstrated production deployments in latency-sensitive, high-throughput environments.
• Strong ML fundamentals: deep familiarity with transformer architectures, fine-tuning, classification, and sequence modeling.
• Python ML stack proficiency: PyTorch or JAX, HuggingFace ecosystem, experiment tracking (W&B, MLflow, or similar).
• Experience building ML feature pipelines and low-latency model serving infrastructure.
• Understanding of LLMs in production (prompt engineering, RAG, tool use, agent orchestration frameworks).
• Security intuition;  you understand adversarial threat models, think like an attacker, and design systems with abuse in mind.

Nice to Have

• Background in cybersecurity, threat intelligence, SOC workflows, or security operations.
• Familiarity with autonomous agent frameworks (LangChain, AutoGPT, Claude tool use, OpenAI Assistants, CrewAI, etc.).
• Contributions to open-source ML tooling or security research publications.
• Experience with eBPF, kernel-level telemetry, or systems instrumentation for behavioral monitoring.

Why CompFly

You will work on one of the most technically novel problems in applied ML today;  real-time behavioral security for autonomous AI agents. This is a high-ownership role where your systems sit in the critical path of enterprise security decisions. We offer competitive compensation, meaningful equity, full benefits, and the rare opportunity to do cutting-edge ML engineering at a mission-driven, early-stage company.

Annual Salary:

$100,000-$200,000 + bonus + equity + benefits.

How to Apply:

E-mail resume at careers@compfly.ai