“Your first customer is probably American”. Heard this yesterday at the @BetaKit meet and it’s kinda sad. As Canadians we should be the first to take a chance on our fellow founders.
Looking for some sponsors to host a mixer for the Toronto ML systems group thursday/friday this week (to gather folks in town for TMLS)! (we have one so far, but will need more to make the event worth people’s time)
Also open to good venue recs!
@adibvafa Was also exploring the same idea for capstone but was thinking about how context rot would affect this. Would be cool if someone ran an experiment.
I built the fastest serving layer scheduler for agent systems (bold claim 😅). AgentIR is a predictive orchestration layer above model servers that schedules agents as workflows, not independent calls.
read the report:
https://t.co/hgYxGw5etU
4 months are the best but if you have a good/supportive team I’d highly recommend doing a 8 monther. The scope’s on a different level.
(Also might apply more to researchy teams)