Research Engineer, Reasoning & Memory - SIML

The System Intelligence and Machine Learning (SIML) Content Understanding teams are seeking a Research Engineer in Reasoning & Memory Systems. You will be working alonside teams that are in charge of operating system wide embeddings, personalized RAG workstreams, tool calling, context compaction / efficiency & memory systems. Projects are focussed on advancing Apple Intelligence capabilities, while working closely across disciplines with our partners in hardware engineering, design and product. \\n\\nSelected references to our prior work (a) , (b) , (c) ;br>
Important attributes expected in the role is fluency in algorithm development (prompt optimization, post training / alignment), and experience with automatic evaluation techniques. The role includes the opportunity to partner with world class system engineers to prototype and incorporate bleeding edge algorithmic innovations in the context of emerging agentic experiences\n\nOther responsibilities include testing and upkeep of training infrastructure, whiled partnering with safety/security teams on emerging robustness challenges while aligning models/agents to production needs. Ability to interface with large scale data infrastructure is a huge plus. \n\nApple has a thriving Machine Learning research community. It is expected that role offers the candidate an opportunity to form a strong network of collaborators across the company, while sharing research progress with senior technical leaders at a regular cadence.

PhD, or MSc in Computer Science/Electrical Engineering, or a related field (mathematics, physics or computer engineering); with a focus on machine learning, or comparable professional experience\nStrong ML and Generative Modeling fundamentals\nProven experience in one of the following: Reinforcement Learning, Multimodal Training, Pre-training / Post-training foundation models \nProficiency in using ML toolkits, e.g., PyTorch\nTrack record of research contributions demonstrated through publications in top-tier conferences, or open source contributions to algorithm

Experience with building & deploying Multimodal-LLMs\nFamiliarity with distributed training and large-scale data infrastructure