Well, indirectly. He took a job at APh Technological Consulting [1], which developed the Intellivision hardware, system software, development tools, and early games for Mattel.
In 1983, APh was working on a next generation Intellivision system that would greatly expand the capabilities of the hardware. One of the big improvements was to be a new graphics chip. The hardware people had prototypes of the new graphics hardware implemented in discrete logic across several wire-wrapped cards in a card cage, and the software people could use those for testing when the hardware guys weren't tinkering with them. Thus, the software people and the hardware people tended to work out of phase.
We software people would get to work, and find notes from the hardware people telling us what changes they had made to the hardware while we were gone...and often these were big changes. They might have eliminated or added color modes, changed the way moving object animation worked, changed the rules on when graphics memory could be accessed, changed command protocols, and other drastic things.
When the hardware people made a big change, it would take must of us hours to get our games working again. By "most of us", I mean "everyone except Hal". Somehow, Hal would bring up his code in his editor, tweak it for a short while, and his game worked again. This was not code written in a high level language, either. This was tight assembly code for an ugly processor.
[1] Or maybe it was APh Corporation rather than APh Technological Consulting. One owned the other, and I forget which was the one the people who worked on the Mattel stuff were actually employed by.