Job Summary
Agentic AI Solution Architect: -
The Agentic AI Solution Architect designs and delivers agentic systems that can be trusted in the enterprise. The role architects the full agent stack: workflows, tools, agents, human gates, orchestration, skills, prompts, memory and context, MCP integration, APIs and events. It builds in evaluation, observability, security, controlled autonomy and AgentOps from the start. It also produces reusable reference implementations and leads strategic pursuits, shaping solutions, winning client trust, and handing off to delivery teams with patterns that scale.
The role is a hands-on, Al-native engineer-architect who uses agent harnesses in daily work, and who can explain trade-offs to both CXOs and engineers.
Architect workflows, tools, agents, human gates, orchestration, skills, prompts, memory/context, MCP integration, APIs/events, evaluation, observability, security, controlled autonomy and AgentOps; build reusable reference implementations and lead strategic pursuits. Mandatory across all roles: practical AI-native engineering; strong modern application development; cloud-native, API, microservices and DevSecOps grounding; hands-on experience with at least one agent harness such as Claude Code, GitHub Copilot, Codex, Kiro or Cursor; enterprise security, governance, evaluation, observability and cost awareness.
Key Responsibilities
A. Agentic Solution Architecture Architect end-to-end agentic solutions: workflows, tools, agents, orchestration, skills, prompts, memory/context, and integration points. Choose the right level of agency per use case: deterministic workflow, single agent with tools, or multi-agent system, and justify the choice on reliability, cost, and risk. • Design orchestration patterns (supervisor/worker, planner/executor, event-driven, graph-based, human-in-the-loop) and select frameworks and runtimes accordingly. B. Tools, Skills, Prompts & Context • Design tool and skill catalogs: clear contracts, schemas, permissions, idempotency, and error semantics. Engineer prompts and system instructions as versioned, testable assets. Architect memory and context management: short-term and long-term memory, retrieval, context-window budgeting, summarization, and state persistence across sessions and agents. C. Integration: MCR APIs & Events • Design and integrate MCP servers and clients to expose enterprise systems, data, and capabilities to agents safely. Architect API and event-driven integration (REST/GraphQL, webhooks, message brokers, async workflows) with enterprise systems of record. • Apply cloud-native, microservices, and resilience patterns (retries, timeouts, circuit breakers, idempotency, compensation) to agent-to-system interaction. D. Controlled Autonomy & Human Gates • Define autonomy levels and guardrails per action type (read, recommend, act with approval, act autonomously). • Design human-in-the-loop gates: approval, escalation, review queues, override, and audit trails, with clear criteria for when a human must intervene. • Establish policies for spend limits, action scopes, kill switches, and safe rollback. E. Evaluation, Observability & AgentOps • Define evaluation approach for agents: task success, trajectory and tool-use correctness, groundedness, safety, latency, and cost; build regression suites. Design observability end to end: traces across agent steps and tool calls, quality and drift monitoring, cost telemetry, and incident triage. Establish AgentOps: versioning ofprompts, skills, tools, and agents; release and rollback processes; runbooks; SLOs; and continuous improvement loops. F. Security, Governance & Responsible AI Threat-model agentic systems: prompt injection, tool misuse, excessive agency, data exfiltration, privilege escalation, supply-chain risk in tools and MCP servers. • Design identity, least-privilege access, secrets handling, tenantisolation, and audit logging for agents and tools. Align solutions with enterprise governance, compliance, and Responsible AI requirements. G. Reference Implementations & Reuse Build reusable reference implementations, accelerators, and templates (agent scaffolds, tool/MCP server templates, orchestration patterns, evaluation harnesses, observability dashboards). • Use Al-native engineering and agent harnesses to build quickly while maintaining production quality. Publish reference architectures, decision guides, and playbooks; mentor engineers and enable delivery teams. H. Strategic Pursuits & Client Leadership Lead technical solutioning for strategic pursuits: discovery, solution shaping, architecture, effort and cost estimates, risk framing, and proposal content. Run client workshops and demos; build proofs of concept that demonstrate value and feasibility. Act as trusted advisor to client technology and business leaders on agentic strategy, adoption roadmaps, and operating models. Partner with Advisory, COE, and delivery teams to turn won pursuits into successful programs
Skill Requirements
Bachelor's or Master's degree in Computer Science, Engineering, or an equivalent field, or equivalent practical experience. • 12+ years in software, solution, or enterprise architecture, with at least 3 years designing and delivering LLM-based or agentic solutions. • Track record of delivering at least one agentic or workflow-automation solution to production, and of supporting or leading client pursuits or pre-sales. Technical Skills (Must Have) 1. Agent architecture: single- and multi-agent design, orchestration patterns, planning, tool use, workflow vs. agent trade-offs. 2. Agentframeworksandruntimes: hands-on with at least one Agent SDK, OpenAI Agents SDK, Semantic Kernel, AutoGen, CrewAI, or custom). 3. Tools, skills&prompts: tool/function-callingdesign, skill packaging, prompt and context engineering, structured outputs. 4. state management, context-window management, long- term memory patterns. 5. MCP & integration: building and consuming MCP servers; APIand event-driven integration; enterprise system connectivity. 6. Controlled autonomy: guardrails, approval workflows, policy enforcement, human-in- the-loop design. 7. Evaluation: agent and workflow evaluation, golden datasets, I I M-as-judge, regression 8. Observability & AgentOps: tracing, monitoring, SLOs, versioning, release management, incident handling for agent systems. 9. Security: threat modeling for agents, identity and least privilege, secure tool execution, prompt-injection defenses. 10. Al-native engineering: practical useofAIacross the development lifecycle; hands-on with at least one agent harness (Claude Code, GitHub Copilot, Codex, Kiro, or Cursor). 11. Modern application development: strongproficiencyin at leastone languagesuchas Python or TypeScript; clean architecture, testing, code review. 12. AWS/Azure/GCP, containers, serverless, service design, event-driven architecture. 13. DevSecOps: CI/CD,infrastructureascode, secrets management, secure SDLC. 14. Enterprise readiness: security, governance, and cost awareness, including token/model cost modeling and FinOps for agent workloads. Behavioral & Professional Skills • Executive presence; able to explain agent architecture, risk, and value to CXOs and to engineers. • Structured problem solving; comfortable with ambiguity in early-stage pursuits. • Consultative, commercially aware mindset; balances ambition with delivery realism. • Leadership, mentoring, and collaboration across Solutions, COE, and Advisory teams
Other Requirements
Role-specific Multi-agent protocols and interoperability (e.g., A2A, agent registries, tool gateways). Workflow and process automation platforms (BPMN engines, RPA, iPaaS) and how they combine with agents. Knowledge graph, GraphRAG, and semantic-layer awareness for agent context. Agent security tooling and guardrail frameworks; red-teaming of agentic systems. Durable execution and workflow engines (Temporal, Step Functions, Durable Functions). • Voice, browser/computer-use, and multimodal agent patterns. Pursuit craft: solution pricing models, outcome-based commercials, proposal writing, competitive positioning. • Domain exposure in BFSI, healthcare/lifesciences, manufacturing, retail, telecom, or public sector. Change management and operating-model design for Al-augmented teams. Common competencies (mandatory across all roles) Practical Al-native engineering and strong modern application development. Cloud-native, API, microservices, and DevSecOps grounding. • Hands-on experience with at least one agent harness (Claude Code, GitHub Copilot, Codex, Kiro, Cursor). • Enterprise security, governance, evaluation, observability, and cost awareness. Certifications (preferred, not mandatory) • Cloud solution architect certifications (AWS, Azure, or GCP). • TOGAF or equivalent architecture certification. • Security credentials relevant to cloud and AI (e.g., CCSP, cloud security specialty).