ML Security Engineer: Real-time Prompt Guardrails

Fortinet Inc.; Sunnyvale (CA); Presencial; USD 150.000 - 183.000; Jornada completa

Ventajas ofrecidas por este puesto de trabajo

Medical insurance

Dental insurance

Vision insurance

Life insurance

Disability insurance

401(k)

Paid holidays

Vacation time

Sick time

Comprehensive leave program

Descripción de la vacante

Fortinet is hiring a Machine Learning Engineer to own detection models behind FortiAIGate, Fortinet's AI security gateway. You will train, evaluate, and optimize guardrail classifiers detecting prompt injection, jailbreak attempts, and sensitive data exposure, and deploy them in production with Triton/vLLM.

Work spans encoder/decoder models, quantization, distillation, and model hardening. You will partner with cross-functional teams to ensure low latency per request and robust evaluation;

Formación

• Strong Python and production PyTorch experience; comfort with Go/Rust/C/C++ for performance-critical paths is valuable.
• Demonstrated experience training, fine-tuning, and evaluating transformer models — encoder classifiers, decoder language models, or both — with Hugging Face Transformers or equivalent.
• Production experience with a modern inference serving system (Triton, vLLM, TensorRT-LLM, TGI), including batching and memory tuning for throughput.
• Practical model optimization: quantization, distillation, pruning, or graph compilation, with a record of maintaining accuracy while reducing latency/memory.
• Sound evaluation instincts — be able to design test sets and reason about precision/recall to avoid false positives blocking legitimate traffic.
• Working knowledge of tokenization, text normalization, and Unicode handling, and how these become an attack surface in a security product.
• Familiarity with containerized deployment (Docker, Kubernetes) and standard MLOps: experiment tracking, model versioning, reproducible training pipelines.
• Ability to deliver on schedule in an Agile environment and communicate effectively across technical and non-technical teams.

Responsabilidades

• Build and train guardrail models. Develop classifiers that detect prompt injection, jailbreak attempts, unsafe content, and sensitive data exposure.
• Design and tune tiered detection cascade to hit accuracy targets within a fixed latency budget.
• Work across encoder and decoder model families; fine-tune encoder-based classifiers and token-level taggers.
• Optimize and serve models inline; quantize, distill, and compile models for GPU appliances and deploy on Triton/vLLM.
• Harden models against evasion; research obfuscation, encoding bypass, and multi-turn attacks; create regression tests.
• Own evaluation and governance detectors; build benchmarks, monitor drift, and maintain multilingual coverage.

Conocimientos

Python

PyTorch

Go

Rust

C/C++

Transformer models

Hugging Face Transformers

Inference serving

Triton Inference Server

vLLM

TensorRT

quantization

distillation

pruning

Docker

Kubernetes

Agile

Herramientas

ONNX Runtime

TensorRT

vLLM

Triton Inference Server

Git/DVC

Descripción del empleo

Fortinet is hiring a Machine Learning Engineer to own detection models behind FortiAIGate, Fortinet's AI security gateway. You will train, evaluate, and optimize guardrail classifiers detecting prompt injection, jailbreak attempts, and sensitive data exposure, and deploy them in production with Triton/vLLM.

Work spans encoder/decoder models, quantization, distillation, and model hardening. You will partner with cross-functional teams to ensure low latency per request and robust evaluation;