ML Engineer - Agentic AI Guardrails

AI/ML

ML Engineer - Agentic AI Guardrails

Why Polygraf AI?

At Polygraf, we envision a future where AI augments human capabilities without compromising safety, privacy, or ethical standards. Trust in our commitment to building this future with you.

About Polygraf AI

Polygraf AI redefines AI security for critical operations. Our proprietary Small Language Model (SLM) technology enables organizations to detect, explain, and mitigate AI risks – from data leakage and compliance violations to deepfakes and synthetic content – using local, explainable, and auditable AI solutions.

We’re award-winning and investor-backed: named ‘Best in Show’ at SXSW 2025 (winning in Enterprise, Smart Data, FinTech & Future of Work), recognized as Top AI & Data Product (2025), Top AI Governance Product and Top AI Content Detection Product (2024) by Products That Count, and ranked a Top Analytics Startup to Watch Globally by Dealroom.

About the Role

Polygraf AI’s core technology is a suite of Small Language Models (SLMs) purpose-built for air-gapped, regulated environments. As agentic AI systems – from coding assistants to fully autonomous pipelines – become the new attack surface, we need someone who can bring our guardrails into that world.

In this role, you will own the integration of Polygraf’s SLM-based guardrails into agentic AI systems and define the protocols that govern how AI agents behave safely within enterprise environments. You’ll work across the full stack: from model evaluation to tool-level integration, producing both production integrations and the internal reference architecture that the rest of the team will build on.

What You'll Do

Integrate Polygraf’s SLMs as guardrail layers into agentic AI environments (e.g., Claude Code, OpenHands, Cursor, AutoGen, LangGraph, custom tool-use pipelines)
Design and author an Agent Safety Protocol – a formalized framework governing how AI agents interact with Polygraf’s guardrails across prompt, tool-call, and output checkpoints
Evaluate agentic workflows for security and compliance risks, and map them to appropriate SLM intervention points
Build and maintain integration adapters, or middleware that make Polygraf’s guardrails plug-and-play for agentic platforms
Collaborate with the product team to translate integration findings into product requirements and documentation
Produce internal benchmarks and evaluation reports on guardrail performance within agentic contexts
Stay current with the fast-moving agentic AI landscape and proactively flag new integration opportunities or risks

You Have

Strong Python skills and hands-on experience with LLM or SLM inference and fine-tuning
Direct experience working with at least one agentic AI framework (LangChain, LangGraph, AutoGen, CrewAI, or similar)
Understanding of how tool-calling, multi-step reasoning, and memory work in agent architectures
Ability to design and document technical protocols or integration specifications
Comfort working in a fast-paced startup environment with minimal hand-holding
Strong written English for technical documentation

Nice to Have

Experience with air-gapped or on-premise ML deployments
Familiarity with compliance frameworks: CMMC, ITAR, HIPAA, or FedRAMP
Prior work in cybersecurity, red-teaming, or adversarial ML
Contributions to open-source agentic AI tools or frameworks
Experience with model quantization or edge deployment (ONNX, llama.cpp, etc.)

As an equal opportunity employer, we highly value diversity and inclusion. We recognize that a diverse team brings unique perspectives, fosters ongoing innovation, and deepens our connection to the global community we serve. If you’re enthusiastic about spearheading the next era of generative AI and leaving a profound impact on how individuals and brands create, we’re eager to have you join our team.

AI/ML