Agentic AI Developer - AI and Data Science Engineer III
Deloitte
- Location
- Onsite (Hyderabad, Telangana · Hyderabad, Telangana · Hyderabad, Telangana · Hyderabad, Telangana)
- Employment
- Full-time
- Level
- Mid Level
About the role
Deloitte Consulting Products (USI) is seeking an AI Engineer to build and ship production-grade agentic AI capabilities for consumer-focused businesses. The role involves designing multi-agent workflows, RAG pipelines, and evaluation systems on Google Cloud Platform to drive personalized experiences and business growth.
Skills
Full job details
Agentic AI Developer - AI and Data Science Engineer III
Bengaluru, Karnataka, India
Chennai, Tamil Nadu, India
Hyderabad, Telangana, India
Pune, Maharashtra, India
Position Summary
AI Engineer, Consultant
Deloitte Consulting Products (USI)
About the Role
Level: Consultant
Are you an AI Engineer who has taken LLM-based applications past the prototype stage and into production? Are you looking for an opportunity to build agents that some of the biggest and most influential consumer businesses in the world rely on every day? If the answer to these questions is yes, we would like to talk with you.
We are looking for an AI Engineer to build and ship the agentic capabilities at the core of our product. This role is deeply hands-on. You will design and deliver AI agents, build the retrieval, memory, and evaluation systems around them, and take them to production on GCP. You will own features end to end — prototype, evaluate, harden, deploy, and then improve them based on how they actually behave with real users and real data.
Key Responsibilities
Drive Ideas to Execution
· Build and ship production AI agents, from single-agent tool use to multi-agent workflows, using LangGraph, LangChain, or similar frameworks.
· Apply the right agent pattern to the problem — ReAct, plan-and-execute, reflection, deep agents, and human-in-the-loop (HITL).
· Design and deliver RAG pipelines: embeddings, vector databases, hybrid search, and reranking.
· Build SQL and dataframe agents that let users interrogate consumer data in natural language.
· Integrate tools and external systems through MCP (Model Context Protocol), tool calling, and structured outputs.
· Manage agent context and memory so agents stay context-aware across steps and sessions, within context window limits.
· Define and run evaluation and observability for LLM features — LLM-as-a-judge, LangSmith and/or Langfuse — and act on what the results show.
· Build and maintain the FastAPI services, data access, and deployment pipelines that carry these capabilities into production.
· Work across multiple LLM providers (OpenAI, Anthropic, Gemini) and own prompt quality, cost, and latency.
Build the Network
· Collaborate with product, data science, and engineering peers to turn consumer problems into shipped AI features.
Qualifications
Required
· 2–3+ years of hands-on AI engineering experience, including deploying LLM-based applications to production.
· Bachelor’s or Master’s degree in Computer Science, Engineering, or a related field.
AI and Agents
- Experience building and shipping AI agents, from single-agent tool use to multi-agent workflows, using LangGraph, LangChain, or similar frameworks.
- Familiarity with agent patterns such as ReAct, plan-and-execute, reflection, deep agents, and human-in-the-loop (HITL).
- Understanding of reasoning concepts and how to implement them, such as explicit reasoning layers (e.g., sequential thinking) versus a model’s native reasoning.
- Experience delivering SQL and/or dataframe agents to production.
- MCP (Model Context Protocol), tool calling, and structured outputs.
- Context management and memory: keeping agents context-aware across steps and sessions, within context window limits.
- RAG: embeddings, vector databases, hybrid search, and reranking.
- Evaluation and observability: LLM-as-a-judge, LangSmith and/or Langfuse.
- Experience with multiple LLM providers (OpenAI, Anthropic, Gemini) and strong prompt engineering skills.
Must have Engineering
- Python, FastAPI, and REST APIs.
- PostgreSQL.
- Docker, Kubernetes (GKE), and GCP, including Vertex AI.
- CI/CD and system design for scalable AI applications.
Preferred
· A2A protocol and other agent frameworks (e.g., OpenAI Agents SDK, CrewAI, Google ADK).
· Agent security and guardrails, including prompt-injection defense.
· Cost and latency optimization (caching, model routing).
· Redis, Kafka/Pub/Sub, Celery, and Elasticsearch.
· Terraform, Helm, and OpenTelemetry.
· Fine-tuning or working with open-source LLMs.
Education
BE/B.Tech/MS/M.Tech degree or equivalent from accredited university
Bachelor’s degree in Computer Science, Information Systems, or another relevant discipline, or equivalent experience.
Location
Bengaluru/Hyderabad/Pune/Chennai/Mumbai/Gurgaon/Kolkata
Shift Timings
11 AM to 8 PM or 2 PM to 11 PM IST as per business requirements
Why This Role Matters
The team
ConvergeCONSUMER™ is a product-driven business that combines differentiated consumer insights with robust prediction and optimization capabilities to help consumer-focused businesses improve decision-making and deliver personalized experiences to drive growth, consumer loyalty, and profitability. We operate with the speed and agility of a startup within the world’s largest professional services firm.
Requisition code: 374375