Looks super useful, will deploy. I've found context window management the killer limiter / skill in enabling many agents flows. Difference between it being magic, and not working. Looking forward to any solutions here.
OctaveWealth [1] YCS12 is also a full stack flat fee 401k.
We provide a very flexible investment menu, with our own Octave Target Risk ETF Portfolios, in addition to Vanguard Mutuals, and fully custom. The Octave portfolios are built inhouse using best of breed Exchange Traded Funds from multiple providers.
re investments, we can include several fund families, including Vanguard and DFA. However most customers prefer our Octave portfolios which are built using best of breed funds from multiple providers, using a sophisticated optimization process.
re craziness, our team already has tons of experience managing billions of dollars at scale. I wrote a large part of the portfolio management system that manages BlackRock's iShares, and my cofounder was responsible for trading $6B/day of flow for large pension funds.
Founder here, happy to take any qs.
We started this because we were sick of percent based fees eroding our investment growth. We also wanted access to more sophisticated investment options which until now have not been available via high end personal wealth managers.
I'd love to have all logs stored from day 1 (ie unlimited retention).
Makes looking for bugs a million times easier, when you don't need to open up archived logs from S3.
Might be an interesting technical challenge to the keep search fast with 100gbs of logs.
Did you consider using views/triggers on the database to migrate to the new API? [1]. Would still have to change all models, though static typing would make this much easier/safer.
I've found sqitch to be a little heavy. It's basically recreating parts of git as tables. Would love to see a lightweight postgres tool that can keep track of a dag of migrations.
Currently using pg-migrator [1]. Seems to have the right balance between power and simplicity. Though numbered migrations can be a little annoying to manage cross team.
They actually do this at Imperial College, London for computer science. Programming 1 is 100% haskell [1]. The idea is to have a level playing field for all students as many may have had prior experience in imperative languages.
Yea, totally agree that memory is at the heart of the problem. Not only in needing a unified view, but also in bandwidth constraints that seem to be the bottleneck in so much code (von neumann bottleneck).
I think a real solution would not just be software only, it would include a hardware component (or tuned to work with a particular chip): a SoC with a few fat cpu cores, many lite core, a GPU, some fpga fabric, all sharing a large cache subsystem and memory.