Agentic AI Developer / Generative A...

Agentic AI Developer / Generative AI Developer - Bengaluru

Deloitte Consulting is seeking an experienced Agentic AI Developer with strong hands-on experience in Generative AI and agent-based systems to join our team in Bengaluru. You will be instrumental in designing, building, and deploying intelligent agent workflows using modern LLMs, RAG architectures, and Azure AI foundation services to solve complex business problems at scale.

The ideal candidate excels in client and stakeholder communication and possesses a proven ability to develop AI agents and Agentic AI solutions. This includes expertise in reasoning, planning, and deploying models on at least one cloud platform, with a preference for Azure. You should also have experience with production deployment of RAG pipelines and custom AI agents.

Key Responsibilities:

• GenAI based development: Demonstrate a strong understanding of Generative AI and Large Language Models (LLMs).
• Agentic AI Frameworks: Possess hands-on experience with Agentic AI frameworks such as CrewAI, LangChain Agents, AutoGen, or Semantic Kernel.
• AI Safety & Security: Apply knowledge of AI model safety, security, and necessary guardrails.
• RAG Architectures: Maintain solid knowledge of RAG architectures, chunking strategies, and retrieval techniques.
• Text Embeddings: Experience with text embeddings using models like OpenAI, Azure OpenAI, or sentence transformers.
• Prompt Engineering: Expertise in prompt engineering techniques including few-shot, chain-of-thought, and structured outputs.
• Model Fine-tuning: Practical experience in model fine-tuning and evaluation.
• Data & Document Processing / Embeddings: Build pipelines for document ingestion, parsing, chunking, and embeddings generation using frameworks like LangChain, Haystack, or custom solutions. Implement vector database solutions (e.g., Pinecone, FAISS, Azure Cognitive Search) for efficient semantic search and RAG implementations.
• RAG Pipelines Development: Design and develop Retrieval-Augmented Generation (RAG) architectures that combine embeddings-based retrieval with LLM-based generation for accurate, context-rich responses. Optimize retrieval strategies for performance, relevance, and cost-effectiveness.
• Prompt Engineering & Prompt Flows: Develop and optimize prompts, chains, and prompt flows for diverse use cases to maximize LLM output accuracy and relevance. Systematically test prompt templates and maintain prompt repositories with versioning and evaluation metrics.
• AI-Ops, Fine-tuning & Model Development: Implement AI-Ops practices for managing LLM workloads, including monitoring, logging, and automated retraining. Fine-tune open-source or proprietary LLMs on domain-specific datasets to enhance performance for targeted tasks. Integrate external data sources with model pipelines to build dynamic and responsive AI solutions.
• Model Versioning & Evaluations: Manage model versioning using MLflow, SageMaker Model Registry, or Azure ML for reproducibility and controlled deployments. Conduct comprehensive model evaluation, including prompt output evaluations, accuracy, relevance scoring, and user feedback analysis. Generate evaluation and accuracy reports for both technical and business stakeholders.
• Collaboration & Documentation: Work closely with data scientists, software engineers, and business analysts to deliver robust Gen-AI solutions. Clearly document architectures, data flows, prompts, and evaluation methodologies for knowledge sharing and compliance.

Required Skills and Experience:

• Generative AI & LLMs: Strong experience with LLMs such as OpenAI GPT, Azure OpenAI, or open-source models like LLaMA, Mistral, Falcon, etc. Knowledge of RAG architectures and vector search solutions for building context-aware applications.
• Embeddings & Vector Databases: Practical experience in generating and managing embeddings for text/document data. Hands-on experience with vector stores such as Pinecone, FAISS, Milvus, or Azure AI Search.
• Prompt Engineering: Expertise in designing and optimizing prompts, chains, and prompt flows to achieve task objectives efficiently.
• AI-Ops & MLOps: Implementation of model versioning, deployment pipelines, monitoring, and automated retraining. Familiarity with CI/CD for AI models, Kubernetes deployments, and containerization.
• Programming & Frameworks: Strong programming skills in Python with libraries/frameworks such as LangChain, Lang Graph, Transformers, SentenceTransformers, PyTorch, or TensorFlow. Experience integrating Gen-AI solutions into web apps, chatbots, or enterprise workflows.
• Model Evaluation & Reporting: Building evaluation pipelines for prompt outputs and model responses, including quantitative and qualitative metrics.
• General Skills: Excellent problem-solving and analytical skills. Strong communication and documentation abilities to explain complex Gen-AI solutions clearly.