About this role
Johnson Controls is hiring a Tech Lead - Data Science in Mumbai, Mahārāshtra, India, to help build a greenfield Agentic AI Platform. The role leads the architecture and delivery of production-grade, autonomous agents that plan, reason, use tools and execute business workflows. It also sets technical direction for GenAI initiatives and mentors junior data scientists. The position offers a flexible hybrid working model.
Responsibilities
- Build and productionize multi-agent systems with planning, memory, tool calling and orchestration. Develop end-to-end RAG pipelines covering ingestion, chunking, embeddings, vector search, re-ranking and grounding; connect agents to enterprise systems and APIs through tool calling, MCP or similar patterns.
- Establish evaluation for agentic workflows, including task-completion measures, hallucination detection, trajectory analysis, LLM-as-judge and A/B testing. Implement prompt-injection defenses, content filtering, role-based tool permissions, human-in-the-loop checkpoints and audit logging.
- Fine-tune or optimize LLMs when needed, balancing model quality, latency and cost. Build classical ML models for classification, regression, forecasting or NLP where useful. Own experiment tracking, versioning, CI/CD, tracing, monitoring and retraining strategies; work with data engineers on data quality and governance. Translate business problems into solutions, present decisions to senior stakeholders and lead code and design reviews.
Required qualifications
- A bachelor's or master's degree in Computer Science, Data Science, Engineering, Mathematics or a related field. The posting states an overall experience range of 8–12 years and specifies 8–10 years of professional data science/ML experience, including at least 3 years building production LLM or GenAI applications.
- Production experience with at least one agentic framework, such as LangGraph, LangChain, AutoGen, CrewAI or Semantic Kernel. Strong knowledge of LLM application patterns, Python, SQL, classical ML, vector databases, Azure AI and data services, and MLOps/LLMOps; experience with TensorFlow or PyTorch. Candidates must be able to address hallucination, prompt injection, data leakage, and cost and latency constraints.
Preferred qualifications include MCP or agent SDK experience; Microsoft AI-102 or DP-100 certification; open-source LLM fine-tuning and serving; agent observability, containers and deployment, PySpark, knowledge graphs or graph RAG; and open-source contributions or published technical content. The posting offers performance-based incentives, certification sponsorship, a learning budget and a technical leadership pathway, but gives no specific salary amount.