AI Engineer Agent & Runtime Enginee...

SecNinjaz Technologies LLP seeks an experienced AI Engineer specializing in Agentic AI & Runtime Engineering to join their team in New Delhi, India.

About the Role:

You will be instrumental in developing agent workflows and runtime services that enable complex cybersecurity tasks. This role bridges advanced software development with domain-specific security expertise, requiring you to build reliable systems capable of planning, tool selection, and multi-step execution within secure environments.

Key Responsibilities & What You Will Work On:

• Agentic Workflow Development: Design agents that autonomously plan tasks, select tools, interpret results, and manage stopping conditions for complex security investigations.
• Runtime Engineering: Implement robust infrastructure including persistent state management, checkpoints, retries, timeouts, cancellation handling, and recovery mechanisms to ensure system reliability during failures or interruptions.
• Tool Integration & Security Tools: Connect agents to external APIs and internal services using structured interfaces like the Model Context Protocol (MCP), ensuring seamless interaction with security testing tools.
• RAG & Memory Systems: Build Retrieval-Augmented Generation (RAG) pipelines, context management systems, and memory mechanisms that provide relevant knowledge bases and preserve investigation history for LLMs.
• Model Integration: Integrate and compare various Large Language Models (LLMs), focusing on private inference solutions and self-hosted environments using model adapters.
• Execution Controls & Safety: Implement strict scope restrictions, approval workflows, permissions management, resource limits, and isolated tool execution to adhere to security protocols.
• Observability & Debugging: Develop comprehensive logging for tool calls, evidence references, failures, latency metrics, and resource usage to ensure agent behavior is transparent and debuggable.
• Product Iteration: Collaborate with data evaluation engineers to diagnose failures, run regression checks, perform rollbacks when necessary, and continuously release improved product capabilities.

Required Skills & Qualifications:

• Technical Proficiency: Strong Python programming skills with extensive experience in APIs, backend services, Git version control, and debugging complex systems.
• AI Agent Expertise: Hands-on building of LLM applications featuring tool calling capabilities, structured outputs (JSON schemas), and robust failure handling strategies.
• Agent Frameworks: Familiarity with frameworks such as LangGraph or equivalent custom implementations for managing agent logic.
• NLP Fundamentals: Deep understanding of RAG architectures, embeddings, context window management, and the inherent limitations of LLM-generated responses in security contexts.
• Data & Systems: Working knowledge of asynchronous execution patterns (e.g., asyncio), database design, state management strategies, and Linux/Docker environments for deployment.

Nice to Have:

• Demonstrated familiarity with VAPT methodologies, web/API security principles, authentication/authorization protocols, or specific security testing tools.
• Experience with Multi-Agent Coordination (MAC), reusable agent skills architecture, and MCP standards.
• Exposure to local model serving technologies like vLLM, GPU inference optimization strategies, and CI/CD pipelines for AI applications.

Educational Background:

A Bachelor's degree in Computer Science, Engineering, or a related field is preferred. Candidates should demonstrate curiosity about cybersecurity domains and a willingness to learn from subject matter experts.