Deloitte—Cyber CST | Assistant Manager | AI Risk Management | Bengaluru |

Job description

Description: Cyber CST | Assistant Manager | AI Risk Management | Bengaluru | • Job requisition ID : 111774 • Location: Bengaluru • Entity: Deloitte Touche Tohmatsu India LLP The team Deloitte helps organizations prevent cyberattacks and protect valuable assets. We believe in being secure, vigilant, and resilient—not only by looking at how to prevent and respond to attacks, but at how to manage cyber risk in a way that allows you to unleash new opportunities. Embed cyber risk at the start of strategy development for more effective management of information and technology risks. Your work profile We are looking for a highly hands-on Senior AI Engineer to design, build and productionize enterprise-grade AI solutions across Generative AI, Agentic AI, RAG/GraphRAG and Machine Learning. This is an engineering-intensive role requiring strong fundamentals in Python, ML/LLM concepts, retrieval architectures, agent orchestration, backend engineering, cloud deployment and AI evaluation. The candidate must be able to go beyond building prototypes and demonstrate experience designing AI systems that operate reliably under real-world constraints including latency, concurrency, security, scalability, cost and failure recovery. Key Skills Required - Design and develop enterprise-grade applications using commercial and open-source LLMs. - Build LLM workflows involving prompt engineering, structured outputs, tool/function calling and model routing. - Understand LLM inference including tokenization, embeddings, attention, context windows, temperature, top-p/top-k sampling and token generation. - Evaluate and select models based on quality, latency, throughput, concurrency, context limits and cost. - Design model abstraction and fallback strategies to avoid dependency on a single LLM provider. - Optimize LLM applications for latency, token utilization and inference cost. - Agentic AI - Design single-agent and multi-agent architectures for complex enterprise workflows. - Determine when agentic AI is appropriate versus deterministic orchestration or traditional software workflows. - Build agent workflows using LangGraph or equivalent orchestration frameworks. - Educational Qualification: Any Graduate BE / B.Tech (Tier 1 / Tier 2) in CS / IT / EC / E, OR MCA from a recognized university