[Remote] Sr Application Security Engineer - AI-First Development

Note: The job is a remote job and is open to candidates in USA. Las Vegas Sands Corp. is seeking a Senior Application Security Engineer specializing in AI-First Development. The role focuses on designing and validating offensive security tooling and adversary-emulation capabilities to identify and remediate exploitable vulnerabilities across various systems, leveraging AI-assisted development tools.ResponsibilitiesDesign, build, and maintain AI agent workflows that produce offensive security tooling, exploit proof-of-concept code, attack automation, and adversary-emulation artifacts from engagement objectives and authorized scopeDecompose engagement objectives and threat scenarios into discrete, verifiable offensive tasks and tooling components that AI agents can execute effectively within defined boundaries and rules of engagementSelect and configure appropriate AI models, agent frameworks, and offensive tooling for each workflow based on blast radius, target sensitivity, operational safety, and cost considerationsConstruct and maintain operational context that provides agents with approved attack techniques, target environment details, rules of engagement, and safety constraints needed to produce correct, in-scope, and consistent outputsContribute to the offensive toolchain, including reusable testing skills, automation hooks, and project memory files that provide persistent context across agent sessions. Authoring of advanced toolchain components may be developed on the jobSystematically capture attack patterns, technique effectiveness, and findings from each engagement and encode them back into shared context, offensive skills, and agent configurations so that subsequent work becomes more reliableParticipate in collaborative refinement sessions to align on engagement objectives, scope, safety constraints, and context packages before agent execution beginsEstablish and maintain rules of engagement, scope boundaries, written authorization, and deconfliction procedures for each engagement, ensuring all offensive activity remains legal, authorized, and safeApply human oversight at governance checkpoints appropriate to the risk level of each workflow, including pre-execution review, in-flight observation, and post-execution auditReview, test, and approve AI-generated offensive tooling, exploit code, and attack automation, ensuring they meet Sands coding standards, operational safety requirements, and rules of engagement before use against any authorized targetVerify that AI-generated exploits and offensive tooling demonstrate genuine, reproducible impact rather than false positives, and reject findings that cannot be reliably validated or that achieve results outside authorized scopePartner with Cyber Security on Threat and Risk Assessments, vulnerability remediation, AI agent governance, and approved tooling decisions, surfacing offensive findings and exploitability signals that inform their reviewsSupport agent observability practices that track engagement activity, finding rates, exploit reproducibility, and coverage across targets and environmentsProduce clear, reproducible engagement deliverables, including technical findings reports, executive summaries, attack-path narratives, and prioritized remediation recommendations tailored to both technical and executive stakeholdersArchitect and deliver shared offensive tooling, exploitation frameworks, and reusable testing harnesses spanning web, API, cloud, and binary targets that the security team consumes, using AI-First methodologies as the primary development approachDefine attack methodologies, engagement playbooks, tooling interfaces, and safety controls that serve as foundational context for agent-driven offensive workCollaborate with cross-functional teams including engineering, QA, Cyber Security, and IT Operations to translate threat scenarios into executable offensive testing workflowsCoordinate with security and engineering teams across global locations to ensure consistency in testing standards, safety controls, and reporting practicesWrite, debug, and refactor exploit code, offensive tooling, and attack automation directly when agent outputs require manual intervention or when developing novel exploitation techniquesEnsure delivered offensive tooling meets enterprise standards for reliability, maintainability, operational safety, observability, and responsible useDesign and execute offensive testing of the organization's AI and LLM-based systems and agents, including prompt injection, jailbreaks, tool and MCP abuse, guardrail evasion, and model and training-data extraction, drawing on frameworks such as the OWASP LLM Top 10 and MITRE ATLASEvaluate emerging AI models, agent frameworks, offensive tooling, and attack techniques to continuously improve testing effectiveness and coverageMentor team members on AI-assisted offensive security practices, context engineering techniques, and verification methodologiesDocument attack patterns, prompt and context libraries, and lessons learned to build institutional knowledgeParticipate in collaborative construction sessions, guiding agent execution in real time and coaching team members on effective offensive tooling and engagement orchestration techniquesPerform job duties in a safe mannerAttend work as scheduled on a consistent and regular basisPerform other related duties as assignedSkillsAt least 21 years of ageProof of authorization to work in the United StatesBachelor's degree in Computer Science, Software Engineering, Cybersecurity, or a related field, or equivalent professional experienceMust be able to obtain and maintain any certification or license, as required by law or policy5+ years of professional software engineering or offensive security experience, including time in senior or lead positions owning the design and delivery of non-trivial security tooling or leading penetration-testing or red-team engagementsDemonstrated daily use, over the past 6 months or more, of at least one modern AI-assisted development tool such as Claude Code, Cursor, GitHub Copilot, or Windsurf, with the ability to speak concretely about effective usage patterns and failure modesStrong foundational knowledge in at least one major programming or scripting ecosystem (such as Python, Go, C/C++, Rust, or PowerShell) and hands-on experience building offensive security tooling or exploits. Familiarity with additional languages and runtimes is a plusExperience assessing or exploiting workloads on at least one major cloud platform (Azure, AWS, or GCP). Azure experience is a plusHands-on experience conducting offensive security work such as penetration testing, red-team engagements, or application, web, and API exploitation, including familiarity with common offensive tooling (such as Burp Suite, Metasploit, or nmap) and adversary frameworks such as MITRE ATT&CKSolid working knowledge of Windows and Linux internals, networking fundamentals, and enterprise identity systems such as Active Directory, including common identity and privilege-escalation attack pathsDemonstrated experience conducting thorough code reviews, identifying defects in both human- and AI-generated outputs, and providing constructive technical feedbackExcellent written and verbal communication skills, with the ability to articulate technical decisions and trade-offs, and to write clear, actionable findings reports for both technical and non-technical stakeholdersStrong interpersonal skills with the ability to communicate effectively and interact appropriately with management, other Team Members and outside contacts of different backgrounds and levels of experiencePractical experience constructing structured context for LLMs, including prompt design, RAG pipelines, context window optimization, project memory files (such as CLAUDE.md or AGENTS.md), and integration with MCP servers. Familiarity with tactical context management techniques such as plan mode, context editing, and multi-session splittingExperience authoring reusable skills, configuring automation hooks, building custom MCP servers, or otherwise assembling agent toolchains that enable repeatable, production-grade offensive security workflowsHands-on experience with vulnerability research, exploit development, fuzzing, or reverse engineering, including container and Kubernetes attack and escape techniquesKnowledge of secure development practices, the OWASP Top 10, adversary frameworks such as MITRE ATT&CK, and threat modeling. Industry certifications such as OSCP, OSEP, OSWE, GXPN, or CRTO. Experience working within a regulated industry such as gaming, finance, healthcare, or hospitality. Understanding of data privacy and responsible AI principlesExperience with command-and-control or adversary-emulation frameworks (such as Cobalt Strike, Sliver, Mythic, or Caldera), purple-team collaboration with detection engineering, or red-team operations against cloud and identity systemsHands-on experience red-teaming AI/ML or LLM-based systems and agentic workflows (prompt injection, jailbreaks, tool and MCP abuse, model and data extraction), including familiarity with the OWASP LLM Top 10 or MITRE ATLAS and tooling such as PyRIT, Garak, or PromptfooCompany OverviewFounded in 1990, Las Vegas Sands is the preeminent developer and operator of world-class integrated resorts that drive valuable business and leisure tourism in the regions where we operate. It was founded in 1990, and is headquartered in Las Vegas, Nevada, USA, with a workforce of 10001+ employees. Its website is https://www.sands.com.