AI Engineer – Consultant (Tax) – Multiple Cities in India

About the role

Deloitte Tax Services India is seeking an experienced AI Engineer to join its USI Tax Technology team. The consultant will design, build, and deploy enterprise‑grade AI and large language model (LLM) solutions that support tax professionals and clients across India.

Key responsibilities

• Design and develop production‑grade LLM‑powered applications and generative AI solutions for complex tax challenges.
• Build and integrate Retrieval‑Augmented Generation (RAG) pipelines, agentic workflows, and prompt‑engineering frameworks into enterprise services.
• Architect scalable AI integrations on Azure, AWS, and GCP, ensuring reliability, performance, and security.
• Develop, train, evaluate, and fine‑tune machine‑learning models using structured and unstructured data, applying supervised, unsupervised, and NLP techniques.
• Own MLOps practices, including model versioning, CI/CD pipelines, performance monitoring, bias detection, data‑drift management, and model retraining.
• Collaborate with data scientists, software engineers, business analysts, and tax experts to translate requirements into maintainable AI solutions.
• Lead technical design discussions, contribute to architectural decisions, conduct code reviews, and document AI architectures, APIs, and deployment processes.
• Champion responsible AI principles, ensuring data privacy, security, and regulatory compliance.

Required profile

• Strong background in AI/ML with hands‑on experience delivering enterprise AI solutions.
• Proven ability to work end‑to‑end on AI projects, from conception to production deployment.
• Excellent collaboration skills to work with cross‑functional teams including tax professionals.
• Commitment to responsible AI practices and regulatory adherence.

Required skills

• Large Language Models (LLM) and Generative AI
• Retrieval‑Augmented Generation (RAG) pipelines
• Agentic workflows and prompt engineering
• Cloud platforms: Azure, AWS, GCP
• Supervised and unsupervised learning
• Natural Language Processing (NLP)
• Fine‑tuning of large language models
• MLOps, CI/CD pipelines, model versioning
• Performance monitoring, bias detection, data‑drift management