Purpose of the Role This role will be responsible for implementing AI & Machine Learning solutions on cloud-based platforms, explore emerging trends in AI, develop proof-of-concepts and engage with internal and external ecosystem to progress the proof of concepts to production. This role would also involve applying engineering principles from classical software development lifecycle to AI & Machine Learning. This role will work with business and IT stakeholders to support a future-state vision in terms of requirements, principles and models in a specific technology, process or function. Key Roles & Responsibilities: AI Engineer enables trusted, enterprise-ready adoption of AI at Deloitte and its clients This role focuses on consuming, integrating, and operationalizing advanced AI models-from Large Language Models (LLMs) to Small Language Models (SLMs) - into secure, governed, and scalable business solutions. Governance by Design: enforce data-handling policies in code (prompt redaction middleware, retrieval allow-lists, per-use-case policies). Prompt/Agent CI/CD: add evaluation gates (answer quality, safety) to pipelines; canary deploys with feature flags; automated rollback on drift. Observability: implement tracing (e.g., request retrieved docs model output tool calls), latency & cost SLOs; alerts on hallucination/safety incidents. Provider Abstraction: wrap OpenAI/Gemini/Azure OpenAI/Vertex behind an interface; capture provider/region, model/version, and quota routing. RAG Ops: schedule index rebuilds, freshness windows, incremental sync, and search quality dashboards with regression alerts. (Complements existing "RAG, cloud, CI/CD, fintech" content.) Develop APIs and microservices to integrate AI with internal/external applications Preferred : Model Lifecycle: manage SLM/LoRA fine-tuning with consented datasets, synthetic augmentation policies, and model registry entries w/ lineage. Partner with business stakeholders to identify and validate use cases Experience and Qualifications Required Seasoned IT software delivery & AI professionals with an experience in hands on development and application architecture. Core AI/ML Skills: o Programming Proficient in Python, PyTorch, TensorFlow, or similar frameworks o Experience with supervised, unsupervised, and reinforcement learning o NLP Expertise: Solid grounding in Natural Language Processing (NLP) concepts - tokenization, embeddings, semantic search, text classification, and summarization. o Generative AI & LLMs: Strong understanding of Large Language Models (LLMs) and Generative AI (GAI), with hands-on experience in LangGraph, LangChain, LlamaIndex, OpenAI APIs and Model Context Protocol (MCP) for building AI agents and conversational systems, Transformers,



and Prompt engineering o Strong understanding of statistics, probability, and model evaluation techniques o RAG & Knowledge Systems: Practical experience with Retrieval-Augmented Generation (RAG) and vector databases like Pinecone, FAISS, or Weaviate. o Applied AI Engineering: Proven track record of designing, developing, deploying, and managing AI/GenAI solutions in production, including post-production optimization and monitoring. Software Engineering o Experience building scalable applications (FastAPI, Flask) o Robust understanding of cloud platforms (AWS/GCP/Azure) o Proficient with Git, Docker, Kubernetes, and CI/CD for ML workflows Hands on Java, Cloud and Kubernetes principles, o Must have experience in architecting & development of applications on cloud infrastructure using different AWS services (or other public cloud like IBM Cloud, Azure, Google Cloud). o Experience with AWS Cloud paradigms like lambda, cloudfront, s3 o Experienced with different forms of architectures like SOA, Microservices, EDA . o Responsive Web applications, Web Services and batch applications development o Other Good to have skills Application development tools and frameworks like maven, ant, check style, PMD, fortify, junit, SONAR, SOAP UI, REST Assured etc. Hands on with Java/J2EE design patterns Good knowledge of Automation testing and Mocking frameworks Good knowledge of DevOps, Build and Deployment best practices and version control systems pg. 1 Fintech Domain Exposure o Experience or familiarity with financial datasets (transactions, risk scoring, KYC, etc.) o Ability to translate business needs into data science solutions Must-Have Skills A. AI Governance & Compliance Define and enforce AI data-handling policies (PII/PCI/GDPR) across prompts, retrieval, logs, and analytics. Implement redaction/masking, tenant isolation, model risk tiers, and provider due diligence. Own evaluation and approval workflows for prompt/model changes, with audit-ready lineage and retention controls. API Consumers: integrate ChatGPT (OpenAI) and Gemini via provider SDKs with fallback logic and request/response schemas. B. Prompt/Agent DevOps & Versioning Treat prompts/agents as code: Git versioning, testable templates, offline/online evals, feature flags, canary & rollback, and automated regression alerts on quality, safety, latency, and cost. Maintain a prompt registry with change history and owners.



Prompt & Response Hygiene: structure prompts with PII minimization, masking, and policy-aware templates; log redaction. PromptOps Basics: track prompt versions in Git, add unit tests (few-shot fixtures), and run small offline eval sets before merges. C. RAG Ops & Vector Governance Operate hybrid search (lexical + vector) with chunking standards, metadata filters, freshness windows, and index rebuild schedules. Enforce document allow-lists, per-tenant isolation, and groundedness thresholdspre-response. Monitor retrieval quality and drift. RAG Basics: implement eval harness for groundedness/faithfulness; try hybrid BM25+vector; document chunking strategies. Vector Stores: basic ops on FAISS/Weaviate/Pinecone with TTL and metadata filters; understand tenant isolation. Compliance Awareness: handle data classification labels in prompts and retrieval; never cache secrets/tokens client-side. (Positions already mention LangChain/vector DBs & cloud-these bullets frame them for enterprise consumption.) D. Provider Abstraction & Protocols Build an abstraction for OpenAI/Gemini/Azure/Vertex with policy-based routing (region, cost, safety tier). Adopt Model Context Protocol (MCP)/standard tool interfaces to integrate internal systems with auditable schemas and scopes. E. SLM/LoRA Fine-tuning Policy Govern data eligibility & consent, synthetic augmentation rules, IP/bias checks, and signed artifacts. Register fine-tuned models with lineage and evaluation evidence before promotion. Good-to-Have Skills Broader understanding of machine learning algorithms (supervised, unsupervised, reinforcement learning). Exposure to multimodal AI (text + vision, speech). Data engineering & analysis skills: ETL pipelines, feature engineering, EDA. Familiarity with MLOps/DevOps (CI/CD pipelines, monitoring, retraining). Understanding of knowledge graphs, embeddings optimization, and enterprise search integration. Strong collaboration and communication skills to work with cross-functional teams and explain AI concepts to stakeholders. Personal Attributes: o Curiosity and passion for solving complex real-world problems with AI o Strong communication, cross functional and collaboration skills, learning to articulate technical concepts clearly. o Proactive, ownership-driven mindset with attention to detail, Takes responsibility for own tasks and code quality with supervision. o Learns to test thoroughly and fix bugs promptly. o Developing a mindset of caring about outcomes (not just output) as guided by seniors. o Adaptability to fast-paced, evolving environments o Follows established processes and uses automation tools, appreciates guidance on designing for future use but generally fixes immediate problems.

📌 AI Engineer - Deloitte (Delhi)
🏢 TRIGENT SOFTWARE PRIVATE
📍 Delhi