Agentic Systems Engineer

The Pakistan Digital Authority (PDA) is looking for an experienced Agentic Systems Engineer to design, develop, and test AI agents tailored for government applications. This role involves combining software engineering skills with retrieval techniques and tool integration to create reliable AI-driven solutions that perform user tasks while adhering to strict permission and oversight protocols. The ideal candidate will work closely with cross-functional teams to ensure these AI agents operate securely and efficiently within defined controls.

Key Responsibilities

- Translate user requirements into effective agent workflows, choosing between single-agent or multi-agent architectures based on task complexity and reliability needs.
- Develop agent applications using frameworks such as LangGraph or model-provider SDKs.
- Integrate AI agents with authorized tools, APIs, and databases, utilizing Model Context Protocol (MCP) where applicable.
- Build retrieval-augmented generation (RAG) pipelines incorporating vector and keyword search methods, ensuring access-aware retrieval and proper source referencing.
- Author, version, and test prompts, tool descriptions, and structured output formats to maintain high-quality interactions.
- Implement least-privilege access controls, validate inputs and outputs, and enforce human approval workflows for high-risk operations.
- Manage agent memory, context length, and sensitive data, ensuring proper retention policies and isolation between users.
- Design task evaluation sets and measure performance metrics such as task success rate, tool-call accuracy, execution steps, latency, and operational cost.
- Detect and resolve issues like incorrect tool calls, infinite loops, and unreliable outputs by applying execution limits, timeouts, and recovery mechanisms.
- Collaborate with MLOps engineers to trace agent workflows, diagnose failures, and support controlled production deployments.

Required Qualifications

- Bachelor’s degree in Computer Science, Software Engineering, Artificial Intelligence, Data Science, or a related field.
- Minimum of 4 years of relevant experience in software engineering or AI application development, including at least 1 year specifically working with large language model (LLM)-based applications.
- Proven hands-on experience delivering AI applications that involve tool calling, retrieval-augmented generation, or agent workflow design.
- Proficiency in Python or TypeScript, API integration, database management, version control (Git), and software testing methodologies.
- Solid understanding of prompt engineering, retrieval systems, agent state and memory management, structured output generation, and application evaluation techniques.
- Ability to implement permission controls and human approval processes, with strong debugging, documentation, and teamwork skills.

Preferred Qualifications

- Experience with LangGraph or similar AI agent frameworks, Model Context Protocol (MCP) integrations, and hybrid search technologies.
- Knowledge of security challenges such as prompt injection, unsafe tool execution, and unauthorized data access in AI systems.
- Familiarity with government workflows, multilingual AI applications, or enterprise systems that require audit trails and human oversight.

This position is classified under Grade-8 with one vacancy available. Candidates should have at least 4 years of relevant professional experience. The application deadline is November 11, 2026.