Backend + Applied ML Engineer (Guardrails)

About the role

We are looking for a Backend and Applied Machine Learning Engineer to own the backend and model components of our Guardrails engine, which detects and mitigates policy violations in AI traffic for enterprise customers. You will train, fine‑tune, and ship small, cost‑efficient language models and extend the policy engine to cover Shadow AI use cases.

Key responsibilities

• Design, train, fine‑tune, evaluate, and ship small language models and classifiers for guardrail detection (e.g., Llama‑class detectors, GLiNER, spaCy).
• Develop and maintain containerized inference micro‑services on Kubernetes (Graviton/ARM64), optimizing cold‑start, caching, latency, and cost.
• Build evaluation pipelines and datasets to measure model quality objectively.
• Extend the guardrail policy engine from detection to enforceable, configurable policies, including new policies for Shadow AI.
• Implement backend plumbing that connects guardrails to worker services, queues, and dual‑database writes.

Required profile

• 3+ years of production backend engineering experience with a typed language (TypeScript/Node, Go, or Python).
• Hands‑on experience fine‑tuning and deploying transformer/SLM models using PyTorch and Hugging Face Transformers.
• Strong understanding of cloud infrastructure, containers, AWS, Docker, and Kubernetes, with ability to evaluate GPU/CPU inference trade‑offs.
• Pragmatic approach to model size, always selecting the smallest model that meets performance criteria.

Required skills

• TypeScript, Node.js, Go, Python
• PyTorch, Hugging Face Transformers, tokenization, quantization, model evaluation
• AWS, Docker, Kubernetes (Graviton/ARM64)
• LLM inference optimization, cost‑per‑event economics
• spaCy, GLiNER, Llama‑class detectors, ClickHouse (optional)