Senior Machine Learning Engineer - LLM Guardrails

Senior ML / AI Safety Engineer

Full Court Defense is building a runtime security layer for AI agents and LLM applications, protecting them from prompt injection, jailbreaks, data exfiltration, unsafe tool use, and agentic attacks in real time.

We’re looking for a senior hands-on ML / NLP engineer with strong AI safety or AI guardrails experience.

What you’ll do

• Build and improve real-time AI guardrails for LLM applications and AI agents.

• Train, fine-tune, distill, and ship transformer-based classifiers for prompt injection, jailbreaks, data leakage, unsafe agent behavior, and abuse detection.

• Use semantic matching, embeddings, vector-store similarity, and classifier ensembles to detect evolving attacks.

• Own the full model lifecycle: dataset creation, annotation strategy, synthetic data generation, fine-tuning, evaluation, deployment, monitoring, and regression testing.

• Red-team your own defenses, find blind spots, generate targeted attack and benign data, and close gaps without increasing false positives.

• Optimize models for production latency, multilingual coverage, and high-throughput inference.

• Translate AI safety, NLP, and security research into production-grade defenses.

What we’re looking for

• 3+ years in applied ML / data science / NLP, ideally with production systems.

• Strong experience with text classification, transformers, embeddings, semantic similarity, or LLM-based detection.

• Hands-on experience with PyTorch, HuggingFace, or similar ML/NLP tooling.

• Experience building AI guardrails, trust & safety systems, abuse detection, content moderation, fraud detection, or security classifiers.

• Strong evaluation mindset: precision/recall tradeoffs, false positives, regression suites, threshold tuning, and benchmark design.

• Ability to work with messy real-world text data, multilingual data, and adversarial examples.

Strong advantage

• Experience with prompt injection, jailbreak detection, RAG security, LLM safety, or AI agent security.

• Experience with LLM distillation, synthetic data generation, automated annotation, or prompt optimization.

• Experience with vector search, semantic matching, Siamese networks, embeddings, clustering, or similarity-based detection.

• Cybersecurity, offensive security, red-team, or AI safety research background.

• ONNX, quantization, latency-sensitive inference, or production model serving.

How to apply

Apply via LinkedIn, or email:

[email protected]