lgsilver

OnFrontiers | Senior Engineer | Hybrid/Remote | NYC or Baltimore

OnFrontiers is hiring a strong senior engineer with experience programming agents to lead development of our agentic intelligence platform for government contractors.

We're looking for a passionate, curious engineer, who wants to work closely with our founding team and contribute as both a technical expert and a member of the product leadership.

In the role, you'll lead development of our indexing services, LLM integrations and APIs, as well as our SaaS platform, as we continue to scale. The role is hands-on but has the potential to grow into management as the team scales.

Core technologies include OpenSearch, Postgres & PGVector, Go, Python and React, all running on AWS.

https://www.onfrontiers.com/job-postings/lead-engineer

Interested, feel free to email me directly: lgsilver[at]onfrontiers.com

A CS degree costs something like 124 yrs salary for someone in a low income nation—and it’s a much longer harder road. I’m not AI’s biggest fan, but arguably, this type of tech actually narrows the gap. Even if it’s expensive.

Let me know when you do! I like the pattern and APIs you've designed for the SDKs—and would probably rely on a managed coordination layer like you've got. But, in order to build confidence in any product like this, we have to know that if something happened to the co, or you went another direction, we could fork the core and continue on.

And the production infra for running isn't even available, just a pared down "development server" via SSPL. This is a long way from OSS.

Agree. This is absolutely deceptive. It's too bad how the OSS moniker is being misused these days...

If you’re supposed to replace modern cartridges every 3-5 shaves, I’ve been doing it wrong. The steel blades in a cartridge easily stay sharp through dozens of shaves. The reason people replace cartridges is because they get clogged with soap and hair. If you rinse them every time, they last significantly longer than single blades.

Not here to give advice, but always hiring and willing to grab virtual coffee if you want to talk about the market or networking. Info is in my bio.

Databricks has been struggling to defend Delta against the fast-moving improvements and widening adoption of Iceberg, championed by two of its major competitors, AWS and Snowflake. This article seems like a bizarre, and maybe even misleading, artifact, given that no one in the industry is comparing Parquet to Delta. They’re weighing Iceberg, which like Delta, can organize and structure groups of parquet (or other format) files…

Agreed. “Built” is definitely the wrong word here.

He went on to say in a slightly devious tone, “I have no ulterior motives here and in no way benefit from this statement or its implications.”

This is a completely naive view of companies and how they work.

An alternative metaphor is this: imagine you got a dog, raised him, lost your sources of income and couldn’t afford to feed him. You kept him anyway out of pride or loyalty or whatever, and watched as he withered away without adequate resources to make him thrive. At the same time, others were looking for dogs and were fully willing to feed them and support them.

When companies don’t do thoughtful adjustments like layoffs, they end up instead in a cycle of slow withering, where the best resources leave and aren’t replaced, and the quality of the workforce slowly decays.

Layoffs may not be “fun” but they’re a necessary part of operating a successful company… and much better than the alternative.

It’s true that if you can deliver to expectations, you could do it, but here’s the thing, growth in a job isn’t linear.

The amount you make as a manager, director, on up grows by more and more. Most company directors or VPs make 2-10x what a passably performing IC makes. Taking two jobs pretty much assures that you won’t grow in either one. Additionally, it only works when you don’t have too many other responsibilities and the demand is where it is for developers.

The minute demand slows down (like it’s doing now) most employers will start pushing for more in-person time from their teams; and have more options for changing out low-performers.

It’s probably one of the least efficient least sustainable ways you could earn enough to retire.

Most SQL dialects accept a positional or a direct reference to the column name. “1” represents the first column created by the select statement, “name”.

You aren’t alone. I’ve noticed a major uptick in these types of editorial changes that remove or obscure the intent of the underlying content. They don’t seem to follow a consistent editorial standard or voice, and are having a noticeable effect on the quality of the HN UX.

Honestly, it’s not a moral high ground. You’re working three jobs—probably 40-50hrs a week at least—more time than you’re getting to spend on your hobbies or your family. Why not do one of those jobs, twice as well, in half the hours, and spend the extra time with your family and hobbies. You’ll be happier. I promise.

There may be IP-related legal consequences for sure, but honestly, if you’re hacking together multiple jobs to make more money, the most obvious consequence is your own happiness and self-satisfaction.

With that setup, I guarantee you aren’t objectively “good” at any of those jobs, even if your dozing managers think you are.

Creating things is about passion, and about thinking deeply about the problems you’re solving. You, like me, probably started coding for fun, because you enjoyed solving problems. Every day you context switch and de/reengage from the problems you’re solving, you miss that. Find a single problem and environment that challenges you, and I guarantee you’ll be happier, more successful, and less liable.

There’s also a product-related argument here as well. In order to start, most products require a lot of scaffolding—-modeling of business entities, general rw apis etc. These are well represented by the big “monolithic” frameworks like Rails or Django. There’s not a lot of value in developing a proprietary distributed paradigm for these. Having this core taken care of and flexible, then allows It’s devs to build more proprietary functionalities as distributed proprietary services. Building this core backbone for access to basic funtionalities, and then distributing the high-value technologies has been the recipe for many things I’ve built.

I've been scraping several real estate sites for years (not sure why exactly) and have millions of listings in bigquery. Could probably pull specific stats for you if you can give a bit more detail on what you're looking for.

Yep. That's totally fair and you're right. Would be interesting for him to compare / analogize that with the lock companies' approach.

Funny that right after he explained how he keeps his family safe by keeping his face and name off the internet, he spent the rest of the video focusing on the ineffectiveness and "downright stupidity" of security by obscurity.

The irony that you have to log into Facebook to read the statement is not lost...

Okay, but this is a function of the medium not of the people. Crafting successful feedback loops has defined great leaders across industries for hundred(s) of years. A bent toward experimentation is not something that software engineers possess. It's a trait that great leaders build into their businesses.

Agree. We consider non-cached requests costly if they extend beyond around $1.50 / million. That said, we do about 2-3% of requests non-cached. As a content network, that means that a million web requests can be as low as $.30 to .40. This does have a lot of benefit to scale, but ~$10 / million seems high. What stops you from putting a ttl on these and caching them for a couple mins. Probably would lower that number substantially.

Really frustrating. I went through this process recently. The data was a couple orders of magnitude bigger and so I tend to agree that maybe just straight to Redshift / Bigquery would probably work best, but here were our steps:

1.) Insure that ingestion / S3 jobs were stabilized (in our case, the legacy were in Informatica, and maintenance took up all the teams' time). We moved to Luigi for this, but Airflow is great too.

2.) Get Presto schemas defined and make Presto the interface for querying / basic pipelining.

3.) Add Mode Analytics or another basic query UI on top for ad-hoc queries. This cleared a massive bottle-neck for our teams because Analysts and data scientists now have direct access to data w/o technical help.

4.) Build "gold" records, for specific sets/types that are valuable, and get them piped from S3 into Redshift/Bigquery (we built a streaming layer for this). This speeds up querying, makes governance easy, and is extremely reliable.

Honestly, the hardest part here was the change management among our legacy teams.. That said, it's incredible how widely this has been embraced now we have it up and working.

I really agree with this advice. I would add that during those quiet conversations the best leads usually say, "we can make this deadline, but we'll need to do x/y/z", where x is increasing resources by a %, y is decreasing scope by some feature(s), and z is increasing time alotted by some days.

That constructive feedback is actionable and adds value. If your manager (or the business) chooses to ignore it, don't be frustrated, but when it's constructive, finite and actionable, you'll be surprised how often they'll make the changes.

Hm. He is the customer. So I think we can safely assume that the customer didn't get a refund for the charges.

Condé Nast | Full-time | NYC, SF, LA

We're hiring data and front-end engineers at Condé Nast. Come work with a great team, cool tech (Storm, Druid, Presto, and Node), and a huge audience. Feel free to email with questions (lgsilver (at) gmail.com)