GenAI Architect

Long Description

Key Responsibilities

Strategy & Discovery
• Partner with business, product, and portfolio leaders to identify high-ROI GenAI and agentic use cases (knowledge work automation, decision support, customer service agents, code assistants, risk monitoring, etc.).
• Run AI readiness assessments: data landscape, governance, model options, risk and regulatory constraints.
• Produce solution blueprints and adoption roadmaps aligned to enterprise architecture and target operating model.
Architecture & Design
• Design end-to-end AI architectures: prompt/flow orchestration, RAG pipelines, multi-agent systems (planner/executor/critic), tool ecosystems (search, DB, APIs), vector stores, guardrails, observability, and CI/CD for ML.
• Select and integrate LLM providers (e.g., Azure OpenAI, Bedrock, Vertex, Anthropic) with model evaluation criteria (accuracy, latency, cost, safety).
• Define data pipelines for embeddings, chunking, metadata, and governed retrieval (role-based access, PII handling, geo-compliance).
• Architect safety & trust: content filtering, PII redaction, policy enforcement, jailbreak protection, and Responsible AI patterns.
• Plan for scalability, performance, and cost (caching, batching, streaming, quantization, distillation, serverless/containerized deployment).
Delivery & Engineering Leadership
• Lead engineers to implement agent frameworks (e.g., LangChain, Semantic Kernel, LlamaIndex, LangGraph) and workflow orchestration (e.g., Airflow, Durable Functions).
• Establish evaluation harnesses: golden sets, rubric scoring, hallucination tests, toxicity/PII metrics, regression suites.
• Drive MLOps/LLOps: versioning of prompts/flows, model registries, monitoring, drift detection, and feedback loops for continuous improvement.
• Ensure integration with enterprise systems (CRM, ERP, data lakes, APIs) and DevSecOps standards.
Governance, Risk & Compliance
• Implement Responsible AI and model risk management: documentation, auditability, exception management.
• Align with regional regulations and industry frameworks (e.g., PDPA, GDPR, financial services guidelines).
• Define human in the loop (HITL) and escalation paths for critical decisions.
Stakeholder Management & Change
• Translate complex AI concepts into business friendly narratives, TCO models, and OKR/KPI frameworks.
• Conduct enablement: playbooks, demos, training, and adoption programs for business units.
• Build vendor and partner relationships; evaluate POCs and coordinate pilots to production.

Long Description

Required Qualifications
• Bachelor’s/Master’s in Computer Science, Data/AI, Engineering, or related field (or equivalent experience).
• 7–12+ years in solution architecture, ML/AI engineering, or platform engineering, with 2–4+ years hands on GenAI/LLM solutions.
• Proven delivery of production GenAI systems (RAG, tool use, agents) at enterprise scale.
• Strong knowledge of:
o LLMs & Embeddings: model families, context management, fine tuning/adapter methods, prompt engineering.
o Agentic AI: planners, executors, memory, tool routing, multi-agent collaboration, safety and oversight.
o Data & Infra: vector DBs (CosmosDB, Pinecone, Redis, PgVector, Azure AI Search), data lakes/warehouses, microservices, APIs, containers (Docker/K8s), serverless.
o Cloud: Azure, AWS, or GCP—identity, networking, secrets, observability, and cost control.
o MLOps/LLOps: model/prompt versioning, A/B testing, monitoring, evaluation pipelines.
• Excellent communication, stakeholder engagement, and consultative problem solving skills.
Preferred (Nice to Have)
• Experience with Semantic Kernel, LangChain, LlamaIndex, LangGraph or custom orchestration libraries.
• Evaluation & Safety tooling: prompt injection detectors, redaction, policy engines.
• Experience with domain compliance (financial services, telco, healthcare, public sector).
• Hands-on with vectorization strategies, multilingual retrieval, and knowledge graph augmentation.
• GenAI UX experience: conversational design, guardrails in UI, user feedback instrumentation.
• Publications, patents, or OSS contributions in GenAI/agent systems.

Core Skills Matrix
Technical
• LLMs (open & closed source), embeddings, RAG, multi-agent design, tool calling.
• Python/TypeScript; API design; orchestration; CI/CD; cloud services; observability.
• Vector databases, chunking strategies, metadata & relevance tuning.
• Security & Responsible AI: content moderation, PII handling, policy controls.
Consulting & Leadership
• Use case discovery, value cases, ROI/TCO modeling.