Energy-Efficient Decentralised Training Frameworks for Large-Scale AI Models on Geo-Distributed Infrastructure

The Opportunity

This is an outstanding opportunity for a highly motivated PhD candidate interested in 𝐞𝐧𝐞𝐫𝐠𝐲-𝐞𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐭 𝐝𝐞𝐜𝐞𝐧𝐭𝐫𝐚𝐥𝐢𝐬𝐞𝐝 𝐀𝐈 𝐭𝐫𝐚𝐢𝐧𝐢𝐧𝐠, 𝐝𝐢𝐬𝐭𝐫𝐢𝐛𝐮𝐭𝐞𝐝 𝐬𝐲𝐬𝐭𝐞𝐦𝐬, and 𝐥𝐚𝐫𝐠𝐞-𝐬𝐜𝐚𝐥𝐞 𝐟𝐨𝐮𝐧𝐝𝐚𝐭𝐢𝐨𝐧 𝐦𝐨𝐝𝐞𝐥𝐬. The successful candidate will be supervised by 𝐃𝐫 𝐌𝐨𝐡𝐚𝐦𝐦𝐚𝐝 𝐆𝐨𝐮𝐝𝐚𝐫𝐳𝐢 at 𝐌𝐨𝐧𝐚𝐬𝐡 𝐔𝐧𝐢𝐯𝐞𝐫𝐬𝐢𝐭𝐲, with co-supervision and support from leading academic and industry experts.

The candidate will join the 𝐅𝐚𝐜𝐮𝐥𝐭𝐲 𝐨𝐟 𝐈𝐧𝐟𝐨𝐫𝐦𝐚𝐭𝐢𝐨𝐧 𝐓𝐞𝐜𝐡𝐧𝐨𝐥𝐨𝐠𝐲 at Monash University and work closely with 𝐏𝐥𝐮𝐫𝐚𝐥𝐢𝐬 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡, the industry partner on this project. The project focuses on developing 𝐞𝐧𝐞𝐫𝐠𝐲-𝐞𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐭 𝐝𝐞𝐜𝐞𝐧𝐭𝐫𝐚𝐥𝐢𝐬𝐞𝐝 𝐨𝐫𝐜𝐡𝐞𝐬𝐭𝐫𝐚𝐭𝐢𝐨𝐧 𝐦𝐞𝐜𝐡𝐚𝐧𝐢𝐬𝐦𝐬 𝐚𝐧𝐝 𝐚𝐥𝐠𝐨𝐫𝐢𝐭𝐡𝐦𝐬 for training large-scale foundation models across 𝐠𝐞𝐨-𝐝𝐢𝐬𝐭𝐫𝐢𝐛𝐮𝐭𝐞𝐝 𝐢𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞.

This research addresses a critical challenge in modern AI: how to train increasingly large models in a more scalable, accessible, and sustainable way. As AI systems continue to grow, centralised training infrastructure creates significant b...