Fortinet
Software Development Engineer (Machine Learning)
$150K–$183K
Sunnyvale, CA · Hybrid · Full-time · Posted Sep 16, 2026
- Visa sponsorship
- Must be authorized to work in the U.S. without sponsorship.
About the job
Own the training, evaluation, optimization, and production serving of machine-learning detectors for FortiAIGate, which identifies threats and policy violations in AI prompts, responses, and tool calls.
What you'll do
- Build and train guardrail models, and tune tiered detection to meet accuracy and latency targets.
- Optimize and serve models in production; harden them against evasion and maintain evaluation, governance, and drift monitoring.
- Develop benchmarks and detection models for sensitive or regulated data and natural-language policy, including multilingual coverage.
What they ask for
- Strong Python and production PyTorch experience.
- Experience training, fine-tuning, and evaluating transformer models with Hugging Face Transformers or equivalent.
- Production experience with a modern inference serving system, including batching and memory tuning.
- Practical model optimization experience and the ability to evaluate precision/recall and deployment-realistic test sets.
Nice to have
- Security or abuse-detection modeling experience, or familiarity with the LLM threat landscape and OWASP Top 10 for LLM Applications.
- CUDA/GPU profiling, multilingual NER or PII detection, synthetic data or active learning, or adversarial ML/LLM security work.
Pay
$150,000-$183,000 USD base salary
Tools
Python, PyTorch, Hugging Face Transformers, ONNX Runtime, TensorRT, Triton Inference Server, vLLM, Docker
AI focus
Train and evaluate models that detect prompt injection, jailbreaks, unsafe content, sensitive data exposure, and policy violations; optimize and serve them under latency constraints and defend against evasion.
Model-written summary · the original posting is authoritative