I was involved in 1981 in decommissioning the Bristol machine, LEO III/70 which was a 326 model (Frank Land comment: the last LEO).  

Prior to decommissioning I had been employed as a systems engineer to provide engineering support for this machine.  I had been trained at Charles House which was a BT (formerly GPO) installation, as was Bristol and Cardiff.  I was unaware that the guys who were training me had been at the forefront of computer science in the 1950s.  Many of the concepts embodied in LEO are to be found in today's PCs - multi-channel DMA for example, though LEO's version of a DMA controller was called an "assembler" - a term guaranteed to cause confusion in conversation with programmers.  LEO III had the world's first multitasking operating system, called rather quaintly "the master routine".  It also had a microprogrammed CPU - a very advanced idea at the time. 

LEO was beautifully made and all those engineers who worked on LEO were very proud to have been connected with the machine and we were very sad when we switched it off for the last time.  I and (I think) six other colleagues were the last generation of engineers recruited by ICL to work on LEO and I joined the company last of all, several weeks after the others in January 1980.  I have to say that much of what I learned during this period has underpinned my entire understanding of computers and it was a marvellous opportunity for a young engineer to have experienced what some have called "the golden age of computing".