domrdy

It has Sonnet 5 as a usable model. Interesting.

For sure true for specialized ones like MedGemma (healthcare). In my testing, the 27b model is at least on the same level as frontier, and in some cases outperforms them. 4B is insanely good too for some lighter workloads. Thanks G for working on this!

Congrats Andrej. Let me know if you are looking for someone to take over Eureka Labs. We were in the same WoW classic guild. domrdy on x. We can duel for it :). Also, I need $10M if any VCs are reading this.

Wow, this is fantastic. Keep it up, Banaxi-Tech, so much more to learn :) Clearly great parenting, too.

Agreed, the way it is implemented is that candidates do the recording after they submit their work.

I agree with you. I’ve failed many live-coding interviews because I start focusing on how the interviewer perceives me instead of the task at hand haha. Hiring managers still want to hear the candidate talk through a technical problem. I built a way for candidates to screen-record their take-home solution using only their microphone and screen share, with as many takes as they want. This lets hiring managers hear how candidates communicate and explain their technical implementations, a skill that matters on teams. But it means managers have to watch those videos. Many of our customers use it and candidates like it, but it still takes extra effort on the candidate’s and hiring managers part, with no guarantee that it will "pay off".

I think certificates are interesting in theory, and there’s an entire industry built around them. AWS does a solid job with its certifications: if you earned one five years ago, it’s still more or less relevant. I'm not sure how certifications would work for proprietary technology.

Labor laws in many places are less “flexible” than those in the US, and they don’t support what you’re proposing, for good reason. I wouldn’t quit my job or uproot my family just to convince a manager I’m worth keeping.

Thank you. I'm not a native speaker, what I was trying to say was, that it depends on the "use-case".

As with everything, it depends. Live coding interviews work. They’re not the best candidate experience, but they work at Meta, Google scale, minimizing false positives better than most other formats. What makes them stressful is the lack of interviewer training and the abstract, puzzle-like nature of the problems, which you can really only solve if you’ve spent time studying (e.g., LeetCode) or you’re fresh out of college or academia.

I’ve worked in the assessment space for 6 years and have seen many hiring processes, from Fortune 10 companies to startups hiring their first engineers. The range of signal required, "how much time can my engineering team spend with a candidate", and how much candidate experience you can get away with, is huge. I’ve also been a candidate myself and failed many live coding interviews. It made me feel terrible about myself. The last time for a role at Ycombinator (the interviewer was super nice).

When I work on my product, I try to view it through the lens of empowering candidates to show their skills and potential. I encourage our customers to use assessments that somewhat resemble on-the-job skills. I don’t like the phrasing “real work” anymore. An assessment shouldn’t be unpaid labor, it should be a way for candidates to demonstrate that they can do the job and handle future work thrown at them, and for hiring managers to feel confident extending what are often very high salaries in tech.

With AI, unfortunately, short take-homes (what I prefer as a candidate, using my own tools and editor) are becoming harder to maintain as a fair signal due to AI assistance. I’ve seen companies move back to onsite, and competitors deploy all kinds of proctoring and invasive monitoring.

The perfect solution, in my view, would be an assessment where the candidate feels relaxed and able to perform at their best, with their own editor and configuration, knowing that every other candidate in the pool has the same constraints in terms of time and tooling. It’s a tough problem to solve. I think about it daily and have not come up with a solution.

Or a 2012 MacMini, $~90 on ebay. Quiet, small form factor, runs Linux.

I'm using Proxmox primarily to host a mix of VMs and containers that serve various purposes around my home and some utility software / local dev & testing servers. HomeAssistant, Plex, Pi-hole, Monica. Backup once a day to Google Drive.

I've found Raspberry Pis to be too unreliable for my homelab needs due to constant crashing and random performance drops. Instead, I bought two used 2012 Mac Minis with 16GB RAM, 1TB on eBay for about $120 each. I installed Proxmox on them, and it’s been a much better experience. It's nice not having to deal with ARM. They look decent and, most importantly, are super quiet. So can even use them in the living room. Absolutely zero issues with this setup—boring hardware wins.

I think they did a pivot to LLM phone calls? I've tried their library the other day and it works quite well. It even has the "interrupt feature" that is being talked about a few threads up. Supports a ton of backends for transcribe/voice/LLM.

I think you only get 100 requests / day during the preview.

For business class, its often less expensive to buy airline miles than buying a cash ticket. Consider Lufthansa and their Miles & More program. You can currently buy 160k miles for $2,010 (1.26¢ per mile). A return business class flight from the West Coast to Germany is 156k miles and ~$100 in fees. The same flight is usually at least $5,000 for the cash ticket. In my experience, Miles and more tickets are superior to cash tickets, as they can be cancelled and changed for free.

There is availability from SF to Munich on the 25th for 78k miles - go enjoy Oktoberfest!

Absolutely, it's not the most inexpensive option out there. My strategy is, that I only initiate a connection once an agreement with an enterprise client is officially signed. This way, I'm not investing upfront without the assurance of a paying customer. I find this approach works quite well for my situation.

I've been using WorkOS for my SaaS products and I'm pretty satisfied with it. It's straightforward to set up Google/Microsoft/MagicLink (free), including staging and production environments. The best part, though, is that it lets my enterprise customers configure their own SAML/OpenID Connect IDPs.

I get charged per connection, so I just pass that cost onto my customers. As a solo developer, I'd have a hard time supporting so many IDPs with a unified API without this.

They did recently hike up the cost per connection, but they're giving a one-year extension at the old rates, which seems fair.

A Tidbyt: https://tidbyt.com/

It's next to my computer, and I use it for notifications and smaller dashboards: https://github.com/tidbyt/community

On another note, isn't it just fantastic that Amazon made a pinky-swear "promise" to not use patient data it acquired with (Alphabet-backed) OneMedical? I mean, what could possibly go wrong with such an ironclad guarantee? It's not like Amazon has a history of exploiting user data for profit or anything. I feel so much better knowing that our medical information is in such trustworthy hands!

https://www.ftc.gov/system/files/ftc_gov/pdf/2210191amazonon...

Apologies. I will add the correct tag next time.

Ali & James discussing the book on the NFX podcast: https://www.nfx.com/post/super-founders/. Also available on Audible: https://www.audible.com/pd/Super-Founders-Audiobook/15491889...

Interesting, I've wondered why you've taken it down! Your project was such a great help dealing with monaco-language-server, and I was really sad when it just vanished off of Github one day.

Very happy Hexadecimal customer. Keep up the good work :)

Hi Andrew - with CodeSubmit we are trying to improve the take-home coding assignment process for both the candidates and hiring managers. The alternative to something like CodeSubmit is in most cases a home grown process that involves a lot of back and forth, emails, public repositories, and spreadsheets. Our customers prefer seeing short real work samples of their candidates instead of leetcode-style algorithmic heavy candidate filters.

I'm happy to walk you through the product and give you a short demo if you're interested, email is in my profile.

Has anyone here managed to hire people from Iran? Our immigration lawyer told us it would take more than a year to even get an appointment at the embassy - I've tried calling multiple times and ultimately gave up. We're in Europe. It sucked because we had a few promising candidates. If you happen to have any special insight / loop holes that you are willing to share with me over email, I would really appreciate it.

Super interesting product. I wonder how they will tackle the plagiarism problem that a lot of project-based products in this space have.

I also think there is a lot of unconscious bias in the technical interview in most companies and not a lot of people like to admit it. We have a similar take on the problem in our own product: https://medium.com/codesubmit/were-all-a-little-biased-a4749...

The trend is definitely going into project-based assessments, at least here in Europe. I think most candidates would prefer that kind of assessment if companies started to invest time developing them. Most of the time they are either a nonsensical mix of brain-teasers or just take way too long to complete.

Is this related to AWS Activate ? Can we still get the credit if we are already part of the Activate program ?

Yes, that sounds reasonable. It's kinda worrying that some popular smart contracts implement "Pause" functions, callable only by one address (usually the creator of the contract).

mayze (http://mayzetheapp.com) | Munich, Germany | iOs Developer & Backend Developer Intern (Python) | Onsite | Full Time

mayze is a premium, members-only dating app. Each profile is carefully selected to keep our community enjoyable and safe. We are launching in January 2017 in Munich.

We are looking to hire highly motivated individuals that are eager to take on responsibility and help us build and shape an early stage company as one of its first employees. Check out our job openings at http://mayzetheapp.com/Jobs and feel free to reach out to me directly at dom[at]mayzetheapp.com

Haha, I wish all telco companies would list their tech stack as part of their feature list on their homepage!