If you're starting at the same Menlo Park company that I work at, for the long term, you may want to consider taking BART down to Fremont and then arranging a r

If you're starting at the same Menlo Park company that I work at, for the long term, you may want to consider taking BART down to Fremont and then arranging a ride in a carpool/van-pool over the Dumbarton bridge: there are tons of folks who live in Fremont/Newark/Union City and who wouldn't mind a +1 for carpool lane. Many of them also have kids so they tend to come a bit earlier than others (most folks come in between 9:45 and 10:45). Also, I've yet to have had a semi-recurring meeting schedule before 11 am (occasionally there are interviews earlier in the morning, however). That said, irrespective of where in Bay Area you live, Ryan is 100% right: the problem isn't the average commute time (which is very easy to keep under 45 minutes -- this is not LA!), it's the outliers. It's a lot like the JVM: good performance on average but then a Concurrent Mark and Sweep failure happens... Also, congratulations! I'm on the data/systems infrastructure as opposed to machine learning side (but folks move around quite a bit and a lot of teams do both ML and systems programming), but let me know if you've any questions.