Software Development Engineer – Machine Learning

Fortinet

. Build and train guardrail models detecting prompt injection, jailbreak attempts, unsafe content, and sensitive data exposure across prompts, responses, and tool-call payloads .

Posted 9/17/2026full-timeSunnyvale • California • United StatesMid-LevelSenior💰 $150,000 - $183,000 per yearWebsite

Core Competencies

Role fit

Core Competencies

Use this summary to align your resume positioning with the role.

Demonstrates expertise in building and optimizing machine learning models, particularly in the context of prompt injection and content safety. Proficient in deploying and tuning models using advanced techniques and tools while ensuring performance and accuracy in production environments.

Highest-signal resume keywords

Python ProgrammingProduction PyTorch ExperienceTransformer Model TrainingModel Optimization TechniquesTriton Inference Server Deployment

ATS Keywords

Tailor your resume

Applicant Tracking System Keywords

Tip: use these terms in your resume and cover letter to boost ATS matches.

Hard Skills

Model Fine-TuningQuantizationDistillationTokenizationText NormalizationMemory TuningPerformance ProfilingAgile MethodologyMultilingual Data ClassificationAdversarial Machine Learning

Soft Skills

Cross-Functional CommunicationTime Management

Tools & Technologies

Hugging Face TransformersDockerKubernetesONNX RuntimeTensorRTVLLMLightGBMXGBoost

Industry Keywords

Prompt InjectionContent SafetyEvasion MethodsOWASP Top 10Security Modeling

Tech Stack

Tools & technologies

DockerKubernetesPythonPyTorchRustC++Go

About the role

Key responsibilities & impact

Build and train guardrail models detecting prompt injection, jailbreak attempts, unsafe content, and sensitive data exposure across prompts, responses, and tool-call payloads
Construct datasets and take models through release
Design and tune tiered detection cascades to meet accuracy targets within fixed per-request latency budgets
Fine-tune encoder-based classifiers and token-level taggers
Adapt small decoder models for semantic judgment
Use distillation to deploy capable, compact models
Quantize, distill, and compile models using ONNX Runtime, TensorRT, INT8, and FP8
Deploy and tune models on Triton Inference Server and vLLM
Configure batching, concurrent model execution, KV-cache, memory, and multi-stage pipelines
Profile and resolve performance bottlenecks
Research evasion methods including obfuscation, encoding bypass, dilution attacks, indirect injection, and multi-turn attacks
Convert bypasses into training data and regression tests
Build production-realistic evaluation benchmarks and suites
Monitor deployed models for drift
Maintain detection models for personal and regulated data and natural-language policy with multilingual coverage

Requirements

What you’ll need

Bachelor's Degree
Must be authorized to work in the U.S. without sponsorship
Strong Python and production PyTorch experience
Comfort with Go/Rust/C/C++ for performance-critical paths is valuable
Experience training, fine-tuning, and evaluating transformer models with Hugging Face Transformers or equivalent
Production experience with Triton, vLLM, TensorRT-LLM, or TGI
Experience with batching and memory tuning for real throughput
Practical model optimization using quantization, distillation, pruning, or graph compilation
Ability to maintain accuracy while reducing latency or memory
Ability to design deployment-realistic test sets and reason about precision/recall
Working knowledge of tokenization, text normalization, and Unicode handling
Familiarity with Docker, Kubernetes, experiment tracking, model versioning, and reproducible training pipelines
Ability to deliver on schedule in an Agile environment and communicate across technical and non-technical teams
Preferred: security or abuse-detection modeling experience
Preferred: familiarity with LLM threats and OWASP Top 10 for LLM Applications
Preferred: LightGBM, XGBoost, NER, PII detection, multilingual data classification, CUDA, GPU profiling, synthetic data generation, active learning, human-in-the-loop labeling, publications, open-source work, or adversarial ML/LLM security CTF or red-team experience

Benefits

Comp & perks

Medical insurance
Dental insurance
Vision insurance
Life insurance
Disability insurance
401(k)
11 paid holidays
Vacation time
Sick time
Comprehensive leave program
Equity program
Bonus eligibility reviewed at hire and annually at the Company’s discretion
Supportive work environment
Competitive Total Rewards package