Intern - AI Agent Developer
Pubblicato il 01-09-2026 - STMicroelectronics Italia in Italia
Your Role Join our MEMS Software Solutions team We are looking for an intern focused on the end‐to‐end development of an AI agent dedicated to technical support for ST MEMS sensors and related software (drivers, libraries, GUIs, configuration tools, documentation, etc.). The goal of the internship is to design and implement an AI assistance system that integrates heterogeneous information sources (technical documentation, FAQs, code examples, application notes) and automation workflows to provide contextual and actionable answers to customers and internal support teams. The Main Activities Will Be * Requirements analysis for the AI agent supporting MEMS sensors (question types, use cases, required integrations) * Architectural design of the AI agent system (workflows, components, integrations with APIs and document repositories) * Integration of AI models (LLMs / language models and retrieval tools) to understand user queries, search for information in technical documentation, generate structured and technically sound answers * Content ingestion and preparation:
collection and organization of datasheets, application notes, user manuals, FAQs, definition of indexing/embedding and retrieval (RAG) strategies * Development of backend workflows and services to orchestrate AI calls, query data sources (databases, file systems, internal APIs), and log requests and responses for later analysis * Development of access interfaces to the agent (e.g., REST APIs, a simple web frontend / GUI or integration into existing tools) * Definition and implementation of tests (functional and regression) to validate the technical correctness of answers, and verify robustness and reliability of workflows and services * Monitoring and continuous improvement of the AI agent: analysis of real interactions, prompt tuning and refinement of retrieval pipelines, performance optimization (latency, cost, answer quality) * Documentation of the developed solutions (architecture, workflows, usage and m