Intern- AI Engineer

Job Description – AI Engineer[Intern] Position - Internship Job Location - Mumbai Qualification -B.Tech/M.tech/MCA [Pursuing or recently graduated] Duration- 3-6 months Stipend- 30,000 per month Role Overview upGrad is building AI-native learning experiences for millions of working professionals globally. As an AI Engineering Intern, you will work directly on LLM-powered products — from prototyping intelligent solutions to improving overall learning experiences. This is a hands-on, builder role for someone who learns fast, ships faster, and is genuinely curious about what AI can do for education. Job Description AI/LLM development Build and experiment with LLM-powered features — including RAG pipelines, AI agents, and prompt engineering workflows — for upGrad's learning and content products. Develop and test prototypes for personalisation, content generation, and learner engagement use cases. Assist in evaluating model outputs for quality, relevance, and safety in an education context. Collaboration & Integration Work closely with product managers, learning scientists, and senior engineers to translate learner problems into AI solution briefs. Integrate AI modules into existing product workflows using APIs and internal tooling. Document experiments, findings, and implementation decisions for cross-functional teams. Research & Iteration Stay current with developments in LLMs, agentic frameworks, and AI in education; propose experiments worth running. Debug, test, and iterate model performance based on user feedback and product metrics. Skills required Strong Python programming skills — you write clean, readable code, not just working code. Hands-on familiarity with LLMs, LangChain, RAG pipelines, or AI agents — through projects, coursework, or personal experimentation. Basic understanding of model evaluation and deployment concepts. Ability to work independently, ask the right questions, and communicate clearly with non-technical stakeholders. A builder's mindset — you default to let me try something before let me wait for a spec.