Job Title: Senior Consultant | Generative AI | Pune | Engineering
What impact will you make?
Every day, your work will make an impact that matters, while you thrive in a dynamic culture of inclusion, collaboration and high performance. As the undisputed leader in professional services, Deloitte is where you will find unrivaled opportunities to succeed and realize your full potential
Deloitte is where you will find unrivaled opportunities to succeed and realize your full potential.
The Team
Deloitte’s Technology & Transformation Technology & Transformation (T&T) - Engineering, AI & Data (EAID) - Intelligent Apps Team focuses on developing, implementing, and managing advanced software solutions that leverage Artificial Intelligence, Generative AI, and data analytics to transform business operations. This team is part of Deloitte Consulting’s Engineering practice, which combines a product-centric mindset with industry-specific insights.
Learn more about Engineering, AI & Data Practice here
Work you’ll do
Role Summary
As an Agentic AI Engineer, you will design, build, and deploy intelligent applications that use large language models, multi-agent orchestration, retrieval-augmented generation, and cloud AI services. You will work in cross-functional delivery pods alongside solution architects, developers, and business analysts, shipping AI capabilities on rapid, iterative cycles.
Relevant Years of Experience 3-5 years
Location – Pune
Education - Bachelor’s or Master’s degree in Computer Science, Engineering, or a related field
Key Responsibilities
1. Build Agentic AI Systems: Design and implement multi-agent workflows using orchestration frameworks (LangGraph, CrewAI, Semantic Kernel, AutoGen). Build agent architectures with tool use, function calling, state management, human-in-the-loop checkpoints, and error recovery.
2. Implement RAG Pipelines: Build end-to-end Retrieval-Augmented Generation systems — document ingestion, chunking, embedding, vector storage, hybrid search, re-ranking, and response synthesis. Implement advanced patterns: agentic RAG, multi-hop reasoning, self-reflective retrieval.
3. Develop on Azure AI Platform: Build on Azure AI Foundry (model catalog, prompt flow, evaluation pipelines), Azure OpenAI Service (GPT-4o, GPT-4.1, o-series), Azure AI Search (custom indexes, semantic ranking, integrated vectorization), Azure Document Intelligence, and Azure AI Content Safety.
4. Implement MCP Integration: Build and consume Model Context Protocol (MCP) servers to enable standardized agent-to-tool communication. Design tool registries that allow agents to discover and invoke capabilities across enterprise systems.
5. Build Observability & Monitoring: Instrument AI systems with comprehensive tracing — agent execution paths, LLM latency, token usage, tool call success rates, and cost tracking. Use LangSmith, LangFuse, Azure Monitor, Application Insights, or OpenTelemetry.
6. Ship Production AI: Deploy and maintain AI systems in production on Azure (Functions, Container Apps, App Service). Build CI/CD pipelines for AI workloads. Monitor model accuracy, drift, and cost. Implement rollback and A/B testing for prompts and models.
7. Build Data Pipelines: Create pipelines that feed AI systems from enterprise data sources (SAP, CRM, databases, APIs, file stores). Implement text-to-SQL and natural language data access for business intelligence use cases.
8. Design API Layers: Build APIs (FastAPI/Flask for Python, ASP.NET Web API for .NET) that expose AI capabilities to frontend applications and downstream services. Handle authentication, rate limiting, streaming responses, and error handling.
9. Ensure AI Quality & Safety: Implement guardrails for every LLM interaction — input/output validation, hallucination detection, confidence scoring, content filtering, PII redaction. Build evaluation harnesses using RAGAS, DeepEval, or custom frameworks.
10. Contribute to Reusable Platform: Extract reusable components from delivered projects — agent tools, prompt templates, RAG modules, data connectors, evaluation harnesses — and contribute them to the practice’s shared component library.
Must-have Skills
- Python or .NET (C#) — 3+ years: Production-grade development in at least one. Clean code, async patterns, error handling, testing, dependency injection. If Python: FastAPI/Flask, pandas, pytest. If .NET: ASP.NET Core, Entity Framework, xUnit. Willingness to work across both is valued.
- LLM Application Development — 1+ year: Has built and deployed at least one production application powered by large language models. Understands prompt engineering, token economics, model selection tradeoffs, structured outputs, function calling, and streaming. Not just API calls — systems that handle edge cases, fallbacks, and cost.
- Agentic Orchestration Frameworks: Hands-on experience with at least one: LangGraph, CrewAI, Semantic Kernel, or AutoGen. Understands agent state machines, tool registration, multi-agent coordination patterns, and critically — when NOT to use agents.
- RAG (Retrieval-Augmented Generation): Has built end-to-end RAG systems in a real project. Understands chunking strategies, embedding model selection, vector store operations, hybrid search, retrieval evaluation, and common production failure modes (context window overflow, irrelevant retrieval, hallucinated citations).
- Azure AI Services: Working knowledge of Azure OpenAI Service and Azure AI Search. Ability to deploy models, configure endpoints, manage API keys, and debug Azure-specific issues. Familiarity with at least one additional Azure AI service (AI Foundry, Document Intelligence, Content Safety, Speech, or Azure ML).
- MCP (Model Context Protocol) — Conceptual: Understands the MCP architecture: servers, clients, tool discovery, and how it standardizes agent-to-tool communication. Hands-on experience building MCP servers is a strong plus.
- Observability — Conceptual: Understands why AI systems need tracing and monitoring beyond standard APM. Familiar with at least one: LangSmith, LangFuse, or equivalent. Can articulate what to monitor in a production LLM application (latency, cost, accuracy, drift).
- SQL — Intermediate to Advanced: Comfortable writing analytical queries, CTEs, window functions, and working with enterprise data schemas. AI systems need data; you need to get it yourself.
- REST API Development: Can build and consume APIs. Understands authentication patterns (OAuth, API keys, managed identity), versioning, error handling, and rate limiting.
- Git & Collaborative Development: Branch-based workflows, pull requests, code reviews. Treats every PR as a quality gate, not a formality.
Education and Certifications
Education
- B.E. / B.Tech / M.Tech / MCA in Computer Science, Information Technology, AI/ML, Data Science, or a related engineering discipline
- Candidates from non-traditional backgrounds with demonstrated AI engineering skills (open-source contributions, shipped products, published work) will be considered
Certifications (any of the following is a plus, not mandatory):
- Azure: AZ-900 (Fundamentals), AI-900 (AI Fundamentals), AZ-204 (Developer Associate), AI-102 (AI Engineer Associate)
- AWS: AWS Certified Machine Learning, AWS AI Practitioner
- Google Cloud: Professional Machine Learning Engineer, Google Cloud Digital Leader
- Vendor-Neutral: DeepLearning.AI certifications (LangChain, LLM specializations), Coursera/edX verified certificates in AI/ML
Other credentials that add value:
- Active GitHub profile with AI/ML projects or contributions to open-source frameworks
- Published technical blog posts or articles on AI engineering topics
- Conference talks or workshop facilitation on AI/ML subjects
- Kaggle competition rankings or notable ML challenge results
Your role as a leader
At Deloitte India, we believe in the importance of leadership at all levels. We expect our people to embrace and live our purpose by challenging themselves to identify issues that are most important for our clients, our people, and for society and make an impact that matters.
In addition to living our purpose, Manager/ Sr.Consultant/Consultant across our organization:
1. Develop high-performing people and teams through challenging and meaningful opportunities
2. Deliver exceptional client service; maximize results and drive high performance from people while fostering collaboration across businesses and borders
3. Influence clients, teams, and individuals positively, leading by example and establishing confident relationships with increasingly senior people
4. Understand key objectives for clients and Deloitte; align people to objectives and set priorities and direction.
5. Acts as a role model, embracing and living our purpose and values, and recognizing others for the impact they make