We are looking for a senior, hands-on Agentic AI Solution (AI/ML) Engineer who can consult, design, and finalize the end-to-end solution architecture and scope of work for an enterprise-grade Agentic AI platform.
Location: DHA Phase 4 Lahore, Pakistan
Timings: 5pm-2am PKT (Fulltime, onsite)
This role is NOT about coding everything.
It is about:
- Asking the right questions
- Defining what should be built vs what should not
- Designing a secure, scalable, cloud-native agentic architecture
- Producing clear solution artifacts that engineering teams can execute on
What You Will Own (Scope of Work)
- Business Discovery & Problem Framing
- Work with stakeholders to:
- Understand current processes, pain points, and automation opportunities
- Identify where agentic AI truly adds value vs simple automation
- Convert ambiguous ideas into:
- Clear use cases
- Agent responsibilities
- Success metrics (accuracy, cost, latency, risk)
Deliverables:
- Problem statements
- Use-case prioritization
- Agent responsibility matrix
- Agentic AI System Design
- Design multi-agent workflows including:
- Task planning agents
- Reasoning agents
- Execution agents
- Validation / guardrail agents
- Decide when to use:
- Deterministic workflows vs agent-driven flows
- Human-in-the-loop vs autonomous execution
Deliverables:
- Agent interaction diagrams
- Decision trees & control flows
- Cloud & AI Architecture (AWS-Centric)
You will design (not just recommend) the architecture using:
Core Stack
- Amazon Bedrock
- Model selection strategy (Claude, Nova, Titan, etc.)
- Prompt orchestration & guardrails
- Amazon Nova
- Agent orchestration & reasoning layers
- AWS Lambda
- Event-driven execution
- Tool calling by agents
- Amazon Textract
- Document ingestion & structured extraction
- Amazon Bedrock Data Automation (BDA)
- Knowledge grounding
- Vectorization & retrieval
- OpenAI models
- Use-case based comparison vs Bedrock models
- Hybrid model strategy (cost, latency, accuracy)
Deliverables:
- High-level architecture diagram
- Data flow & security model
- Model selection rationale
- Tooling, Integrations & Data Strategy
- Define how agents will:
- Call internal tools
- Invoke APIs
- Query databases
- Handle documents & unstructured data
- Design RAG vs Agentic Retrieval strategy
- Define:
- Prompt versioning
- Memory (short-term vs long-term)
- Context boundaries
Deliverables:
- Tool invocation strategy
- Data & memory architecture
- Integration map
- Governance, Security & Risk Controls
- Define:
- Role-based access
- Prompt & output guardrails
- Audit logging
- Cost controls
- Address:
- Hallucination risks
- Data leakage
- Model drift
Deliverables:
- AI governance checklist
- Risk mitigation plan
- Scope Definition & Delivery Blueprint
This is the most critical output.
You will:
- Break the solution into phases
- Phase 1: MVP
- Phase 2: Scale
- Phase 3: Autonomy & optimization
- Define:
- In-scope vs out-of-scope
- Team roles required (Dev, MLOps, Cloud, QA)
- Time & effort estimates (high-level)
Deliverables:
- Final Scope of Work (SoW)
- Phased roadmap
- Delivery-ready architecture pack
Required Profile (Non-Negotiable)
Background
- 3+ years experience across:
- Solution Architecture
- AI/ML systems
- Automation platforms
- Has designed systems, not just implemented them
AI & Agentic Expertise
- Proven experience with:
- LLM-based systems
- Multi-agent orchestration
- Prompt engineering at scale
- Strong understanding of:
- When NOT to use agents
- Trade-offs between agents, workflows, and APIs
Cloud & AWS
- Deep hands-on knowledge of:
- AWS (Lambda, IAM, S3, VPC)
- Amazon Bedrock ecosystem
- Event-driven architectures
- Able to reason about cost, latency, and scalability
Business & Communication
- Can:
- Translate business problems into technical solutions
- Push back diplomatically on unrealistic expectations
- Communicate clearly with executives and engineers
Nice to Have (Strong Plus)
- Previous consulting or pre-sales architecture experience
- Experience designing AI CoE or platform teams
- Familiarity with:
- LangGraph / LangChain
- Agent frameworks
- Enterprise RPA + AI convergence
- Experience in regulated or enterprise environments
