Intellectsoft is a software development company delivering innovative solutions since 2007. We operate across North America, Latin America, the Nordic region, the UK, and Europe. We specialize in industries like Fintech, Healthcare, EdTech, Construction, Hospitality, and more, partnering with startups, mid-sized businesses, and Fortune 500 companies to drive innovation and scalability. Our clients include Jaguar Motors, Universal Pictures, Harley-Davidson, and many more where our teams are making daily impactTogether, our team delivers solutions that make a difference. Learn more at www.intellectsoft.net

Our client is establishing a dedicated research capability focused at the critical intersection of cybersecurity and artificial intelligence. This role owns end-to-end AI security research problems across the entire modern AI stack - encompassing generative and agentic systems, LLM copilots, machine learning models, training/data pipelines, and enterprise AI governance. The core objective is turning open security challenges into adversarially testable hypotheses and developing exploits and defensive controls that generalize across models and enterprise environments

Requirements

• Experience: 3+ years in security research, offensive security, ML/AI security, detection engineering, or a related discipline.
• Engineering Stack: Strong Python development skills for building research-grade tooling, async services, CLIs, SDKs, and automated reporting pipelines.
• AI & AppSec Fundamentals: Strong foundations in application security, threat modeling, and failure modes specific to autonomous and transformer-based systems.
• Attack Surface Fluency: Hands-on expertise in at least one major AI attack surface (LLM apps, agentic/tool-using workflows, or ML training/inference pipelines).
• Security Domains: Demonstrated experience in vulnerability analysis, detection engineering, adversarial ML, policy-as-code/authorization, red-teaming, or secure SDLC practices.
• GenAI / LLM Architecture: Solid grasp of LLMs, retrieval-augmented generation (RAG), prompt injection, and tool calling.
• Research Judgment: Ability to design experiments with held-out evaluations and statistically rigorous metrics; ability to distinguish exploitable vulnerabilities from theoretical edge cases, eliminate evaluation test leakage, and verify genuine generalization.
• CVE & Literature Translation: Capacity to evaluate cutting-edge security literature, advisories, and CVE disclosures and convert them into operational exploits and defensive mechanisms.
• Tooling Familiarity: Working familiarity with representative tools across categories, including:
• Languages & ML: Python, PyTorch/TensorFlow, Hugging Face Transformers.
• Agent Frameworks: LangChain, LangGraph, LlamaIndex, MCP, OpenAI/Anthropic SDKs.
• AI Red-Teaming & Guardrails: Garak, PyRIT, promptfoo, Giskard, ART, NeMo Guardrails, Guardrails AI, LLM Guard, Rebuff.
• AppSec & Scanners: Burp Suite, OWASP ZAP, Semgrep, Nuclei, ModelScan, picklescan.
• Policy & Observability: OPA/Cedar, Docker, Kubernetes, Terraform, cloud providers (AWS/Azure/GCP), and tracing tools (OpenTelemetry, LangSmith, Langfuse).

Nice to have

• Dual experience spanning both offensive red-teaming and defensive engineering (detection, policy enforcement, governance).
• Practical experience working with agentic architectures, MCP ecosystems, and multi-agent frameworks.
• Applied knowledge of AI governance and risk frameworks (NIST AI RMF, ISO 42001, EU AI Act, MITRE ATLAS, OWASP LLM & ASI Top 10).
• Experience implementing policy-as-code (Cedar, OPA), applied cryptography (digital signatures, hash-chained audit logs), or CI/CD security quality gates.
• Proven track record in cross-framework, cross-dataset, or real-world validation of security controls.
• Published security research, CVE disclosures, security advisories, or competitive CTF / red-teaming accomplishments.

Responsibilities

• Harness & Orchestration Architecture: Design and build core agent harnesses, covering function calling, tool interfaces, context assembly, retrieval integration, subagent delegation, retry logic, and failure handling.
• Prototype to Production Delivery: Take agentic systems from initial concept to scalable production, managing deployment, monitoring, cost, latency, and operational support against live enterprise client data.
• Evaluation Infrastructure: Build and maintain rigorous evaluation suites, task benchmarks, scoring rubrics, and regression tests using tools like Langfuse, Pydantic Logfire, or custom SDKs to drive architectural decisions.
• System & Tool Integration: Integrate client enterprise systems, data sources, and internal APIs as accessible agent tools, including building and managing Model Context Protocol (MCP) servers.
• Solution Consulting & Requirements Translation: Translate ambiguous client requirements into robust agent architectures, clearly advising stakeholders when an agentic pattern is—or is not—the right solution.
• Stakeholder Communication: Interface directly with client stakeholders and articulate technical trade-offs clearly to non-technical audiences.
• Tooling Transfer & Documentation: Contribute reusable harness components to shared internal tooling and document design decisions to meet audit, validation, and handover standards.

Benefits

• Employment-based cooperation
• Comprehensive insurance for you and your family (health, life, and accident)
• Paid PTO policy (vacation, sick leaves, and public holidays)
• Tech equipment provided
• Udemy courses, workshops, trainings & expert knowledge-sharing
• Flexible hours & work setup