bbaumgar

I’m curious to hear what you’re excited about in signal processing?

Maybe I'm not following why this is undesirable behavior? Did the therapy session work?

Would $99 + gratitude be enough to motivate you to take that time? I got serious value out of this compendium [0] and have been looking for more.

Contact info is in my profile if interested.

[0] https://elidourado.com/blog/notes-on-technology-2020s/

Beanstalk Farms (W18) | Full-time Onsite | Senior Software Engineer, Director of Research, and more [0] | Washington DC | https://beanstalk.farm

We're growing the world's appetite for fresh produce in local vertical farms. We are expanding our production and are launching several product lines in the next 9 months.

Software at Beanstalk means an interesting mix of fields: logistical web apps, plant-growth-focused statistics, embedded systems, and whole-farm systems engineering. We are a team of multidisciplinary engineers who pride ourselves on pragmatism.

Our web stack is Postgres + Postgraphile + React, hosted in Kubernetes. Our robotics stack is Python + Arduino deployed via Balena to RPI's and ClearCore's.

Shoot me an email at brent.baumgartner@beanstalk.farm if this sounds interesting. Even if you're not a perfect fit for any of our open positions, we're happy to make up a role for the right engineer or scientist.

[0] All our open positions: https://boards.greenhouse.io/beanstalk

Between idom[0], streamlit[1], anvil[2], and PyWebIO this space is moving quickly.

While it's a little rough around the edges, idom has worked great for us. I look forward to trying out PyWebIO at some point - it targets a different but valid use case.

[0] idom-docs.herokuapp.com/ [1] https://streamlit.io/ [2] https://anvil.works/

Beanstalk Farms (W18) | Full-time | Senior Software Engineer, Director of Research, and more [0] | Washington DC | https://beanstalk.farm

We grow seeds from around the world in farms around the corner. To address the growing demand for fresh produce, we create fully automated vertical farms. We are expanding our production and are launching several product lines in the next 12 months.

Software at Beanstalk touches an interesting mix of fields: logistical web apps, plant-growth-focused statistics, embedded systems, and whole-farm systems engineering. We are a team of multidisciplinary engineers who pride ourselves on pragmatism.

Our web stack is Postgres + Postgraphile + React, hosted in Kubernetes. Our robotics stack is Python + Arduino deployed via Balena to RPI's and ClearCore's.

Shoot me an email at brent.baumgartner@beanstalk.farm if this sounds interesting. Even if you're not a perfect fit for any of our open positions, we're happy to make up a role for the right engineer or scientist.

[0] All our open positions: https://boards.greenhouse.io/beanstalk

They have a "Shop" button on their homepage that lets anyone buy their product for installation on their own car. I had a coworker purchase Comma.ai + a specific make/model solely for its Comma.ai compatibility last year. Average people are getting in on this, for the better or (likely) worse.

There is no built in admin UI but it's about a 15 minute task to connect up to Forest Admin, which plays like a dream.

How can you apply this to debugging useEffect issues?

We are in the process of transitioning off it because of those same Chinese origins... probably justified but it's a painful process.

I use Roam [0] with Neuracache [1] to get a similar workflow to what you're describing. My notes are hierarchical and searchable w/ Roam. If I want to create an Anki entry, I add #flashcard anywhere in the note. I use secondary tags and links to further categorize these notes. The next time I import my Roam db to Neuracache all #flashcard entries are synced.

[0] https://roamresearch.com/ [1] https://neuracache.com/

Do you think that the free market worked as desired in the case of Equifax?

This only applies during December, but https://adventofcode.com/ is a fun experience! Most puzzles range from 30 minutes to 3 hours from start to finish.

We make a large, enterprise-targeted app at my startup. Fullstory has accelerated our search for product-market fit by allowing us to make some significant discoveries in how users use our app.

We could have made all these discoveries by properly instrumenting our code with our own tracking framework. That said, Fullstory provided us a big bang for our buck in [time spent : value provided]; half-day implementation led to all the usage info we could need.

We've gone as far as creating a recurring "Game Tape Friday" meeting where the whole development team sits down and watches 5-10 user sessions together to record bugs and possible UX improvements.

We're not a small website so this anecdote may be of limited value to you, but I would personally give up syntax highlighting before I would give up Fullstory.

This looks interesting! A couple questions:

How does this compare to CoreML? What are the advantages of running models on-device vs in the cloud via an API?

They are exposing a vulnerable and naive class of investors to a higher-risk asset class in an arguably deceitful way. If their 3% account was SPIC protected (and therefore low-to-no risk), this would be a totally different conversation.

I'm curious - how hard have you looked? There are lists with clojure success stories [0] [1] [2] and quite a few companies hiring clojure devs [0] [3] [4].

Anecdotally, there are 5 startups within a 1 mile radius of my office (in Charlottesville, VA) whose primary stack is clojure.

[0] https://clojure.org/community/success_stories

[1] https://news.ycombinator.com/item?id=14302762&__s=r8nvw64mp3...

[2] https://github.com/clojure/clojurescript/wiki/Companies-Usin...

[3] https://www.indeed.com/q-Clojure-jobs.html

[4] http://hnhiring.me/ (search "Clojure", top right)

Good catch!

Fantastic typography feedback, thanks.

Awesome, thanks for the report. I had someone report that before and couldn't reproduce. Looks like I should have tested in base Firefox, not just Firefox Developer Edition :)

I'm about 90% through a rework of my personal site, https://www.brentbaum.com/

Would love feedback on what works / doesn't work design-wise.

Care to explain why you believe this was a mistake? Opinions of IBM's future performance aside, even if he were right we wouldn't know until years from now - value-based investing takes a while to pan out.

It wasn't hyphenated before.

The new context api isn't meant to replace stateless components and recompose [0], it's meant to be a more sound context API that lives almost exclusively in library code, replacing usage of the old context + (maybe) 5% more use cases.

[0] https://twitter.com/dan_abramov/status/984510107923963904

This is alibaba's react component framework. It's extremely high quality, but unfortunately most the english is translated from Chinese, so awkward grammar is abound. I suggest looking past that.

Always more to learn, but I've been happy with the results so far.

My rule when I was unhappy with my rates was simple:

1. Double my rates.

2. If the client mentions my rates, but still agrees, then I haven't increased them enough. Go back to step 1.

3. If the client says no because of my rates, I move on (this can be a good sign - if no one is turned away because your prices are too high, you don't have too high of prices). If enough potential clients say no because of my rates (50% lower conversion rate is break even when you've doubled your rates, so # of clients must more than half), then I lower my rates. I quickly climbed from $20/hour to my current rate using this strategy.

There's a nice effect here where higher rates suggest higher quality, which in turn attracts the right types of clients and makes them more willing to pay. I've also found that I want to give them $X/hour quality work, so I'm more focused and the quality increases as a result. Lots of happy feedback loops.

For your specific situation - do you have many repeat customers? One trick that worked well for me is to price my first project with a client at an introductory rate 20% lower than what I actually want to earn (positioned as a "first project discount"). When that project goes well and they come back for more work, the introductory rate doesn't apply, and you've earned yourself a 20%/hour bonus.

I've used two strategies that have worked well.

First, I followed the advice of people lower in the thread and got embedded in two niche communities (private schools + small business owners in my local area). Second, given that in the past I wasn't a talented networker (that's changing, albeit over years, not weeks) I partnered with two other developers who were skilled networkers and observed how they generated leads and closed sales. The combo of [exposure to potential clientele (1) + knowledge of how to talk to and close those clientele] has been effective.

Another note: all my biggest clients have come through relationships - personal friendships and past customers of software I've created. It's a patient game, for sure, but after a couple years of slowly but steadily completing projects and doing the right things, I'm booked out 6-8 months and have a sustainable pace of work.

It sounds like the issue here is more an issue of work conditions and rates, rather than a problem with the actual work.

A counter-anecdote: A couple years ago I decided to freelance on the side, but went the other way - I charged triple my day rate, and set a simple rule: no stress allowed. If the client had tight deadlines or a stressful personality, I didn't pursue them.

It's been a massively fun few years, and I've been able to pad my bank account while working when and if I want. It was hard to turn down some great opportunities that would have been stressful, but the biggest enemy isn't free time: it's burnout.

High rates and optimizing for lack of stress don't guarantee lack of burn out, but they go a long way.

It seems this is an argument for greater rewards for whistleblowers in other industries - not one for decreasing financial whistleblower rewards.

This doesn't seem to teach "really good UX" - it instead teaches UX that directly drives business value (upsells, conversion rates, etc). While that's certainly valuable, I would call that growth hacking design rather than product design.