EY - Solution Architect - Generative/Agentic AI (3-8 yrs)

Job Summary :

We are seeking a passionate and highly skilled Solution Architect Generative AI & Agentic AI to design, develop, and deploy intelligent systems leveraging large language models (LLMs), multiagent frameworks, and enterprise AI integration patterns. This role will play a key part in building next-generation AI applications that autonomously reason, act, and learn across complex business workflows.

As a Generative & Agentic AI Solution Architect you will own the end-to-end architecture for enterprise-grade GenAI solutions and multi-agent systems. You'll translate business objectives into scalable, compliant, and observable AI capabilities spanning data pipelines, foundation-model ops, Retrieval-Augmented Generation (RAG), and autonomous agent orchestration. You will partner with product owners, data scientists, platform engineers, security, and business stakeholders to accelerate AI-driven transformation while safeguarding performance, ethics, and regulatory compliance.

Key Responsibilities :

1. Solution Development :

- Design and implement applications using LLMs (e.g., OpenAI, Mistral, Claude, Llama)

- Build autonomous multi-agent systems using frameworks like AutoGen, LangGraph, CrewAI, or AgentOps

- Develop scalable backend services and orchestration pipelines using Python/Node.js integrated with AI APIs

- Lead discovery workshops, create MVP backlogs, and convert PoCs into production-ready solutions following MLOps / LLMOps best practices

- Oversee model selection, finetuning, prompt-engineering, evaluation, and continuous monitoring

2. Architecture & Engineering :

- Translate business requirements into agentic workflows with reasoning, memory, and tool usage

- Integrate GenAI and agents with enterprise systems (e.g., SAP, ServiceNow, Salesforce) via APIs and RAG pipelines

- Work with vector databases (e.g., Pinecone, FAISS, Weaviate) for semantic search and long-term memory

- Define reference architectures for GenAI (LLMs, Diffusion, SSMs) and agentic patterns (task-decomposition, planner-executor, tool-calling)

- Design secure micro-service and event-driven topologies on cloud / hybrid infra (AWS Bedrock, Azure OpenAI, Google Vertex, private GPUs)

Required Technical Competencies :

- Generative AI : Hands-on with LLMs (GPT-4o, Claude-3, Gemini, Llama-3), Diffusion, and audio-/vision models; experience with HF Transformers, LangChain / LlamaIndex

- Agentic Frameworks : Experience designing goal-oriented agents using AutoGen, CrewAI, Semantic Kernel, or custom planners; familiarity with tool-calling, memory management, and AI schedulers

- LLMOps / MLOps : CI/CD for model artifacts, feature stores, vector DBs (Pinecone, Weaviate, FAISS), model gateways, and policy engines (Guardrails, Azure AI Safety)

- Cloud & DevSecOps : Terraform, Kubernetes, Docker, serverless, GPU orchestration, secret management, SAST/DAST integration

- Programming : Python (primary), TypeScript/Node, Bash; strong design-pattern discipline and code-review leadership

- Build replaceable but working proof- of- conceptsCLI, Streamlit, or VS Code Jupyterso stakeholders can touch" the idea

- Write production-grade Python/TypeScript to orchestrate agents with AutoGen / CrewAI / Semantic Kernel

- Hook agents to real systems (SAP BAPIs, Salesforce APIs, REST/GraphQL) and manage auth tokens & secrets

- Build CI/CD (GitHub Actions, Terraform, Helm) that auto-tests prompts, pushes Docker images, and provisions GPUs

- Instrument tracing (OpenTelemetry) and set up dashboards for tokens, latency, cost per call

Education :

- Bachelor's or master's degree in computer science, Data Science, Artificial Intelligence, Engineering, or a related field

Note : Working from office Mandatory