Telecommunications Generative AI Foundation Model Development Engineer

This role sits within SoftBank’s LTM AI Development team, which researches and applies the Large Telecom Model, a generative AI foundation model designed for telecommunications operations. The team uses network data, operational knowledge, domain datasets, existing foundation models, and GPU infrastructure to improve automation and autonomy across mobile-network design, management, and monitoring.

The engineer will implement continued pre-training and fine-tuning methods such as SFT and RLHF, optimize training recipes, and design and clean large telecommunications datasets from base-station data and operational logs. The role also covers training pipelines, automated evaluation frameworks and benchmarks, technical investigation of current models and learning methods, and PoC development based on papers and open-source repositories. Collaboration with application teams and operational departments is part of translating business issues into model requirements.

Applicants must have implementation experience in machine learning or deep learning using Python, and practical or research experience with generative AI or LLM training, continued pre-training, fine-tuning, or evaluation. They must also be able to read English papers, technical documentation, and open-source repositories and apply the findings in code, and use AI-assisted development tools such as GitHub Copilot effectively.