I lived in DC for a couple of years and I felt exactly the same. I try and carry that same mentality to all museums now, paid or unpaid. No research, no expectations, wander around and let yourself be surprised, leave when you’re bored. It’s a lot of fun.
Although it is hard to do when you’ve paid 50 bucks for a ticket and you know there is “just one more” ultra-famous piece on the other side of the museum and when will I really be back in (Amsterdam, Paris, etc)?
Are there platforms that make such training more streamlined? Say I have some definition of success for a given problem and it’s data how do I go about generating said RL model as fast and easily as possible?
We're working on an OSS industrial-grade version of this at TensorZero but there's a long way to go. I think the easiest out of the box solution today is probably OpenAI RFT but that's a partial solve with substantial vendor lock-in.
Typically speaking an LLM is the code driving the control flow and the MCP servers are kind of dumb API endpoints (find_flights, search_hotels, etc) say for a travel MCP.
With your product, how is the LLM made aware of the underlying data store in a more useful way than “func search(query)”?
It seems to be that if you could expose some precomputed API structure into the MCP for a given data store then the LLM could reason more effectively about the data rather than throwing search queries into the void and hoping for the best?
From what I have gathered their main differentiator is taking the approach of assigning each discrete data point its own "entity" definition that is independent but can be extended for each data provider.
So since its all represented by entities, you could treat them like any other vectorised data in your vector data store and use vector search.
It's a nice technique, but probably tricky if they ever venture into encapsulating endpoints in realtime for rapidly changing b2c applications (ratelimits/cronjob latency)
You should go. I was all set to go to Caltech or the Ivies and turned them down for mostly, though not entirely, financial reasons and went to a state school instead. Worked out mostly fine but much harder to get foot in the door and the opportunities and experiences you get in a place like MIT last a lifetime.
If you don't go, many doors will become closed or so much harder to get into that it's basically equivalent. Want to join a think tank, become a quant on Wallstreet, easily raise VC money at 23? These are all much easier with the credibility a MIT degree brings.
"Normal" degreed people essentially live a life where everybody assumes you are stupid until proven otherwise. MIT people are assumed to be smart until proven otherwise. That is a tremendous advantage in almost all contexts.
Not to mention the quality of your peers and the education itself. Don't pass up the opportunity to really challenge yourself and see what you can do. Sure, it's just undergrad and you aren't really solving anything of note but just being surrounded by the leaders of your field is inspiring and will push you to be your best.
Sample size 1 over here but especially MIT I would go. MIT is big and diverse enough that you can go as technical as you want, or study business, or…?
Anyway my 2 cents from a 30 something that has been around the block.
> Want to ... become a quant on Wallstreet...? These are all much easier with the credibility a MIT degree brings.
Do enough hard math EE at MIT and becoming a quant as your first job is a common initial career step. The money you'll make will then help you do whatever you want going forward.
Not sure how this would fit in with the OP's desire to get a CS Ph.D.; on the science track it's very easy to fall off, I'd expect CS to have more flexibility but I don't know.
Just wait until all software becomes a streamed H.265 video stream. That's certainly the future once latency allows for it. No piracy, infinite subscriptions, total control. But if you don't like it, don't buy it. That's life.
It's a different approach, coupled to automating the desktop office ecosystem, whereas we're coupled to web-apps, and web APIs alone.
Secondly, it is more complicated, a bit more like Leapwork, whereas we're targeting Zapier-level complexity.
Axiom is already too complex for many people (it's why we mainly target these no-code Zapier types). We've seen every marginal % increase in complexity reduces the number of people who can build bots significantly.
Essentially, each RPA product has chosen a power vs ease-of-use trade-off for different segments. We're fixated on the Zapier / Airtable people, not the traditional desktop RPA people, whom I think Microsoft are targetting.
I'm sure they will face the piper at some point, but you don't see people in Copenhagen living on the streets or in squalor. Probably because China makes all of their stuff and immigrants build all of their houses, but still.
Although it is hard to do when you’ve paid 50 bucks for a ticket and you know there is “just one more” ultra-famous piece on the other side of the museum and when will I really be back in (Amsterdam, Paris, etc)?