My father, who worked as a market research analyst, said that the trick to the job was "point out the obvious before anyone else does".
Web sites and web content are trending towards being enshittified until it just barely becomes worth wading through all of the shit to get to what you actually wanted to consume.
Python has a much more mature ecosystem than Rust, especially for AI/ML stuff. I ran into a rust crate that purported to do a certain ML algorithm but did not do it correctly. I managed to write a replacement with Claude though.
I do think enforcing correctness at the type system level is a good idea for AI, which is why I often choose languages like C# and Rust over Python. However, for some things Python is definitely the correct tool for the job.
Using either Calibri or Times New Roman makes it look like you did not put any thought into your brand and chose the default in Microsoft Word. The State department probably has certain constraints (i.e. they likely have to choose one of the fonts that ships with Microsoft Word, and possibly a subset of that that also ships on macOS), but they could definitely choose better than the default.
I find the narrow serif typefaces such as Century Schoolbook a bit harder to read than ones with more normal spacing, and I think the US government should optimize for legibility and accessibility over style in routine communications. Palatino or Garamond would probably be my choices.
I don't think I was specific enough on what kind of structured data. The idea is that it extracts information from the text/HTML content of emails (e.g. a flight itinerary from an airline booking email or an ingredient list from a recipe) using AI.
Since you already have a method for reaching into folks Microsoft 365 inboxes and such, you could probably train an LLM to extract arbitrary data based on a user's prompt quite quickly though.
The first is a preventive maintenance and calibration tracker (https://pmcal.net) that was born out of my day job as an engineer in small business manufacturing.
The second is an AI engine for pulling structured data out of incoming email (either via IMAP on your email server or via SES). If you think of the engine that powers TripIt, they had to write about 10,000 different ingestors for each airline and hotel and travel booking site. With a structured output AI, the need to write specific ingestors goes away.
I discovered I liked business planning, strategy, and marketing pretty early in my technical career.
I also discovered that one of the thing business people bring to a company are connections, especially if they have already worked in a vertical, and few people want to talk to a bright kid about their problems.
The quality of store management seems to affect a lot when it comes to Walmart. I have been to some Walmart stores where I have never seen any unfolded clothing. I have been to some Walmart stores where I have never seen any folded clothing.
An organization as large as Walmart has the wherewithal to put together an effective charging network if they decide it is a priority. One of the nice things about being first-party is that they will likely have a preventive maintenance program and emergency maintenance program for the chargers, and some of that can be done by the company electrician.
I have a Tesla, and due to various factors (apartment living, other family members with Teslas), I almost exclusively use public chargers. I have some opinions of what amenities I want at a charger (bathroom, cold diet coke), but in general the chargers at gas stations, grocery stores, and Walmarts are the best for me. Restaurants and malls are actually not great amenities because that's often not the mood you are in when you arrive at the charger.
Thank you. My day job is as an engineer in manufacturing so I felt like I had a unique opportunity to "build what I know" and have seen people use and enjoy.
It's a worthwhile project to build yourself. If nothing else I found out that I definitely do not like the date-fns library in JavaScript. I built it using AWS Amplify, and although I like that it scales to zero, but I think there are too many gotchas to Amplify, and especially DynamoDB, for a startup app that you want to move quickly on. I wrote up one of the major ones after I got really frustrated. [1]
Like I said in my original post, I am trying to figure out how to get it in front of the right people (who are less likely to be on HN). I have kind of decided that the B2C sales experience is not great unless you get a critical mass; my experience doing sales in manufacturing is working the booth at trade shows, talking to people about engineering, and using our process tools to develop a solution to the customer's problem. The more scattered "compete for attention" advertising/promotion sales model doesn't seem great unless you have a lot of money behind it.
I'm rambling, but if anyone likes this or feels it needs a certain feature, feel free to reach out. If you're in Boston / Providence I'll happily grab a drink with you.
I wrote a small business preventive maintenance and calibration tracker (https://pmcal.net) as a side project.
A few manufacturing companies that I have a close relationship with are using it and love it, but I have kind of hit a wall with other growth avenues (Google Ads, organic promotion on the web).
I have been thinking of marketing directly to ISO 9001 auditors, because “can you get email reminders” is a question they have asked at multiple companies I have worked at. I feel like cold mailing them something branded (e.g. notepads) might work, but I am not sure how much money I want to spend on it if it doesn’t and it’s also a bit nerve-wracking to put myself out there like that.
I played with OpenHands for a few days (using gpt-4o since I already had an OpenAI account). I found it to be decent at writing new code, but then it had a hard time making changes when there was a lot of repetitive code (in a TypeScript / React project that I had it create with vite).
One of the interesting things about OpenHands is that you can see what the AI is doing in the terminal window where you launched it. Since it can't really load the whole codebase into its context window, it does a lot of greping files, showing 10 lines on either side of the match, and then doing a search and replace based on this. This is pretty similar to what a human might do: attempt to identify the relevant function and change it.
I think I might have better luck with a simpler project, e.g. a Sinatra or Flask app where each route is relatively self-contained. I might give it or Cursor another try in the future when the tech has progressed a bit.
I heard a story of a fishing boat in the eastern US that was "fighting a fish" for miles but could never get any traction. When another fisherman looked at their chart, he noted that they were dragged miles in a straight line towards Europe, and said "You caught a sub". The submariners don't care, they probably find it funny.
Our area measurement application did not require that tight a tolerance (we were estimating yield on broken material). If I needed that tight a tolerance, I could have gotten proof coins from the mint, or potentially switched to using a real calibration standard like a gauge block.
I built a computer vision device that used the top-down area of a penny as a calibration standard. Coins are useful, easy-to-get items that have relatively tight manufacturing tolerances.
Generally in order to ticket an itinerary, there needs to be at least one flight marketed (e.g. with the airline's flight number) by the "plating carrier" whose ticket stock the flights are issued on.
I can't buy an itinerary consisting of just BA238 on aa.com but I can buy AA6981 which is its codeshare. I can also buy an itinerary where I fly AA on BOS-JFK and connect to BA JFK-LHR, because there's at least one AA-marketed flight on the ticket.
The marketing carrier also can affect how the operating carrier gets paid -- codeshares can have different inventory which allows airlines that are partners but not super close to hold back inventory for themselves.
I'm not as familiar with foil resistors, but if you apply the same rules as thin film I would expect that you could trim this with a normal meter (the ± is 0.1 ohms) and the major cost driver is laser trim time and the space cost of some extra geometry on the resistor to support the tight tolerance (i.e. more fine trim features than standard).
I believe Vishay's ±0.2ppm/°C TCR is a materials science process specific to one of the companies they own, so that is also a reason they can charge quite high prices.
If you don't need that kind of TCR (i.e. your part is not going to space) the price should go down considerably for a thin film NiCr resistor (5 ppm - 25 ppm). There is actually a lot more direct sales and custom design volume than I would have expected when I started in this industry, so what you see on Digi-Key is not the entire market.
It depends on the component and the company/process. Laser trim time for thin film is a significant cost-driver, so if possible you want to aim for a specific value and reject or bin-sort the rest out. My company only makes 1% tolerance resistors by laser trimming.
Thin-film resistor design engineer here! It's dependent on value and geometry -- if you order a 0.5 ohm resistor the meters on our trimming lasers only go down to 20 mΩ and you're getting a 5% part at best.
Web sites and web content are trending towards being enshittified until it just barely becomes worth wading through all of the shit to get to what you actually wanted to consume.