Assistant Manager | Data Scientist | Bengaluru | Sustainability & Emerging Assurance

Summary: Design, build, and deploy AI agents and intelligent automation solutions using Azure OpenAI and AWS Bedrock. The role involves integrating AI with existing infrastructure, developing reusable components, and ensuring secure, governed production deployments while transferring knowledge to internal teams.

Assistant Manager | Data Scientist | Bengaluru | Sustainability & Emerging Assurance
• Job requisition ID : 115431
• Location: Bengaluru
• Entity: Deloitte Touche Tohmatsu India LLP

We are seeking an experienced, hands-on AI Implementation Engineer with 10–12+ years of overall engineering experience to design, build, deploy and operationalise AI agents and intelligent automation solutions within CLD/Sys, working alongside the existing automation team.

The engineer will implement AI agents, AI Copilots, Retrieval-Augmented Generation (RAG) solutions and agent-driven workflows using enterprise-approved platforms such as Azure OpenAI and AWS Bedrock. These solutions will integrate with existing infrastructure tools, automation frameworks, APIs and enterprise knowledge sources.

The role will own practical engineering delivery from use-case identification and solution design through coding, integration, evaluation, production deployment and continuous improvement. A key responsibility is to build sustainable internal capability through reusable components, documented implementation patterns and hands-on knowledge transfer.

Working closely with Architecture, Security and Infrastructure teams, the engineer will deliver secure, governed and supportable solutions.

This is a senior, hands-on implementation role, not primarily a management, advisory, research or standalone enterprise AI platform ownership position.

Key Responsibilities

AI Agent Engineering & Implementation

• Design, develop and deploy AI agents, operational assistants and AI-enabled automation workflows for CLD/Sys.
• Implement agent capabilities including tool calling, planning, state management, contextual memory, workflow orchestration and human-in-the-loop approvals.
• Build single-agent solutions and, where justified, multi-agent workflows.
• Implement AI solutions using Azure OpenAI, AWS Bedrock and appropriate agent orchestration frameworks.
• Develop RAG solutions connected to approved runbooks, technical documentation, knowledge repositories and operational records.
• Evaluate agent behaviour, retrieval quality, output accuracy and execution reliability.
• Take solutions beyond proof of concept into production-ready, monitored and maintainable services.

Integration with Existing Automation

• Connect AI agents to existing Python scripts, REST APIs, infrastructure automation, monitoring and service management platforms.
• Reuse and enhance established automation frameworks rather than create duplicate tooling or a parallel delivery model.
• Combine AI-driven analysis and recommendations with deterministic automation for controlled execution.
• Implement approval gates, execution validation, failure handling and rollback mechanisms with infrastructure specialists.
• Prioritise practical use cases that reduce manual effort and improve operational reliability.

Building Internal AI Capability

• Develop reusable agent components, tool connectors, retrieval services, evaluation suites and deployment templates.
• Establish shared engineering patterns for agent development, testing, deployment and support.
• Work alongside automation engineers through pairing, code reviews and practical implementation workshops.
• Document architecture, source code, configurations, dependencies, limitations and support procedures.
• Ensure the internal team can independently maintain, extend and scale the solutions delivered.

Software Engineering & Platform Integration

• Develop scalable Python applications, backend services and REST APIs.
• Build reusable software components supporting AI applications and infrastructure workflows.
• Implement automated testing, secure coding and version control across code, prompts and agent configurations.
• Contribute to technical design reviews and engineering standards.
• Deploy private/local LLM environments where required and approved, working with platform specialists.

DevOps, Security & Operational Reliability

• Build and maintain CI/CD pipelines for reliable delivery of AI applications and agent workflows.
• Use infrastructure automation and container platforms to support repeatable deployment.
• Implement least-privilege access, secrets management, data protection and auditable tool execution.
• Apply safeguards against prompt injection, unsupported outputs and unauthorised actions.
• Require human approval for sensitive or production-impacting activities.
• Implement monitoring, logging and alerting for agent quality, execution success, latency and consumption costs.
• Support production readiness, troubleshooting and stabilisation of delivered solutions alongside existing operations teams.

Required Skills & Qualifications

Core Engineering Experience

• 10–12+ years of overall experience in software engineering, platform engineering or infrastructure automation, with strong hands-on implementation expertise.
• Strong Python development, REST API and enterprise integration skills.
• Experience with automated testing, software design patterns, secure coding and maintainable application development.
• Minimum 2 years of hands-on experience delivering Generative AI, Agentic AI or enterprise AI applications.
• Demonstrable experience building and deploying AI agents beyond standalone demonstrations or chat interfaces.
• Ability to independently deliver production-ready solutions and transfer implementation knowledge to internal engineers.

Agentic AI & Generative AI

Practical experience with:

• AI agents, tool/function calling and workflow orchestration.
• RAG, embeddings, semantic search and knowledge retrieval.
• Prompt design, agent state management and contextual memory.
• AI evaluation, guardrails and hallucination mitigation.
• Human-in-the-loop approval and controlled execution.
• Integration of AI applications with operational systems

Technology Stack

Strong implementation experience in at least one of Azure OpenAI or AWS Bedrock is required. Experience across both is preferred. Expertise in every technology below is not expected.

AI platforms and model services

Azure OpenAI, AWS Bedrock; exposure to Azure AI Foundry and enterprise-approved model APIs

Agent development and orchestration

LangGraph, LangChain, Semantic Kernel, AutoGen, CrewAI or equivalent

RAG and knowledge integration

LlamaIndex, embedding models, semantic search and document processing

Vector storage and retrieval

Azure AI Search, PostgreSQL with pgvector, Qdrant, Milvus, Weaviate, ChromaDB or equivalent

Software development

Python, FastAPI, REST APIs and enterprise SDKs

Infrastructure automation

Ansible, Terraform and existing operational scripts and APIs

Deployment platforms

AWS and/or Azure, Docker and Kubernetes

CI/CD

Azure DevOps, GitLab CI or GitHub Actions

Security and observability

IAM, Vault or equivalent secrets management, audit logging, monitoring and agent evaluation