Sr. SW Engineer - AI Security

Visa•Ashburn, VA

1d•$118,700 - $183,600•Hybrid

About The Position

Visa's Cyber Analytics and AI Innovation organization builds the security capabilities that let the enterprise adopt Generative AI responsibly and safely. This is an innovation-oriented team: our charter evolves as GenAI threats evolve, and our core service is a growing suite of guardrail services that protects AI applications, LLMs, agents, tools, and data interactions from AI-specific threats. As a Senior Software Engineer, GenAI Security, you'll be a hands-on individual contributor who owns problems end-to-end — from technical design and prototyping through production deployment, monitoring, and support. Because this is an innovation org, you won't be boxed into one feature lane: your work may span detection engines, agent and tool security, evaluation frameworks, or platform integration as priorities shift, and you'll often be handed a problem with real latitude in how you solve it rather than a fixed spec.

Requirements

2+ years of relevant work experience and a Bachelors degree, OR 5+ years of relevant work experience
Professional software-development experience building backend services, APIs, platforms, or distributed applications.
Experience developing production software using Python, Java, Go, or another modern programming language.
Experience with cybersecurity, secure software development, application security, cloud security, AI security, or a closely related discipline.

Nice To Haves

Advanced degree (M.S. or Ph.D.) in computer science, applied math, or a related field.
Strong proficiency in Python and at least one additional language such as Java, Go, or Rust, with solid grounding in distributed-systems fundamentals — software architecture, design patterns, data structures, concurrency, API design, and performance optimization — applied to highly available RESTful, asynchronous, or event-driven services.
Hands-on experience with Generative AI application architectures, including LLM APIs, embeddings, agentic workflows, tool calling, model gateways, and orchestration frameworks such as LangChain, LangGraph, Semantic Kernel, AutoGen, or LlamaIndex, along with familiarity with MCP, AI-agent tools, plug-ins, or similar emerging integration patterns.
Understanding of GenAI-specific security risks (prompt injection, jailbreaking, sensitive-data exposure, insecure output handling, unauthorized tool use, data exfiltration) and hands-on experience implementing controls such as input validation, output filtering, policy enforcement, access control, audit logging, rate limiting, and secrets management.
Knowledge of secure software-development lifecycle practices — threat modeling, vulnerability management, and application-security principles — applied in day-to-day engineering.
Experience developing evaluation frameworks or test suites for LLM applications, including adversarial or red-team testing, regression testing, and false-positive/false-negative analysis.
Experience with cloud-native deployment and operations — Docker, Kubernetes, CI/CD, infrastructure automation, and observability platforms (such as Splunk, Elasticsearch, or Grafana) for metrics, logging, tracing, alerting, and production troubleshooting.
Familiarity with GPU-backed inference, CUDA, NVIDIA technologies, or model-serving frameworks is beneficial but not required.
Ability to evaluate new technologies objectively against security, quality, latency, scalability, maintainability, and operational-readiness criteria, and to work independently on ambiguous, complex assignments.
Clear written and verbal communication skills, including the ability to explain technical and security tradeoffs to different audiences, and experience collaborating effectively in Agile, cross-functional environments spanning engineering, cybersecurity, infrastructure, research, and product.

Responsibilities

Build and operate GenAI security systems: design, build, deploy, and maintain production services and APIs that detect and mitigate risks such as prompt injection, jailbreak attempts, sensitive-data exposure, and unauthorized or unexpected tool/agent behavior — spanning LLMs, agents, tools, APIs, MCP servers, plugins, and enterprise data sources. Apply deterministic, policy-driven controls where predictable enforcement matters, and model-based approaches where judgment is needed.
Measure and improve: build automated evaluation, testing, and feedback loops, plus production observability (logging, metrics, tracing, alerting, dashboards), so detection quality, latency, and false-positive/negative rates stay visible and auditable — then use that telemetry to close gaps.
Integrate and scale: containerize and deploy services with Docker and Kubernetes, and integrate GenAI security capabilities into enterprise applications, model gateways, developer workflows, and cloud or on-premises platforms.
Practice secure engineering: apply threat modeling, secure coding, dependency and secrets management, code review, and vulnerability remediation; participate in architecture and design reviews and document decisions, interfaces, and operational requirements clearly.
Explore and influence: evaluate emerging GenAI security technologies and frameworks through structured proofs of concept, provide technical guidance to other engineers, and help shape the team's engineering standards and practices.
Collaborate and operate: work with engineers, researchers, architects, product managers, and infrastructure teams to translate emerging AI-security risks into product capabilities, and support production releases, incident analysis, and defect resolution.