maxprogram

Mashgin (YC W15) | ML/CV, Full-Stack, Frontend | Mountain View | ONSITE, VISA | Full-time

Mashgin makes self-checkout kiosks that use Computer Vision, 3D Reconstruction and Deep Learning to scan many items at once without needing barcodes, reducing checkout time by up to 10x. We are live across the United States and expanding quickly.

Our founding team consists of ex Facebook, Microsoft, Bell Labs engineers with vast experience in CV and AI. We’re all hackers at heart and love to build and iterate on interesting projects.

Looking for engineering roles across the spectrum: from frontend to deep learning. Check out or jobs page here: http://mashgin.com/jobs.html

Why Mashgin?

* Get to work on a working, real-world application of deep learning. Our product is currently used by thousands of people a day and will continue to grow exponentially. People already love what we do, but we're only at the very beginning of the improvement curve... 95% of our work remains to be done.

* We have a small but growing team, where every engineer can and will have a huge impact.

* We're in a big, but mostly hidden industry with a huge amount of potential. There has been little innovation in decades and we’re completely recreating the experience.

The market isn't vegetarians/vegans. Their ultimate goal is reduction of environmental damage and greenhouse gases. Why try to sell to someone who already doesn't eat meat? The company is mission-driven and it wouldn't contribute to their cause.

The market is also too small. Reminds me of the apocryphal story of the 2 shoe salesmen in Africa: 1 says bad news, no one wears shoes. The other says good news, no one wears shoes.

Poor Charlie's Almanac -- can't beat Charlie Munger when it comes to explaining how the world works.

Fooled by Randomness, Black Swan, Antifragile -- Nassim Taleb reviles lots of new ways to think, first in finance, then everything in later books.

The Origin of Wealth -- Similar to Antifragile with a lot of mental models packed in on many different subjects: economics, business, biology, ...

The Design of Everyday Things -- the bible of design. Read it to know why everyday frustrations with tech are probably not your fault. His book Emotional Design is a good compliment.

The Essential Drucker -- "essential" reading for anyone in management or scaling a startup.

History, and why the world is the way it is today:

Sapiens, Yuval Noah Harari

Guns, Germs, and Steel, Jared Diamond

The Birth of Plenty, William Bernstein

They Made America, Harold Evans -- fantastic history book with each chapter telling the detailed story of a businessperson or inventor in U.S. history

Mashgin Inc. (YC W15) | Computer vision engineer | Palo Alto, CA | Full-time onsite

At Mashgin, we apply advanced computer vision to real world problems, starting with retail checkout. Our kiosk uses computer vision to scan multiple items at once without barcodes or RFID, reducing checkout time by 10x. But that’s just the beginning -- we’re building the future of machine vision for all businesses.

We’re looking for someone with serious engineering skills that can help us fulfill the promise of computer vision in the real world. You should have experience writing vision algorithms in C++. Bonus points if you’ve worked with 3D cameras, multi-cam scene reconstruction, or machine learning. As the first non-founding engineer you’ll get the chance to shape the foundation of Mashgin.

We’re a team of 3 with previous experience in machine learning at Facebook and Microsoft, and computer vision for humanoid robotics at Bell Labs and Toyota. We’re well funded from top investors following YC W15, and are working hard on getting our machines production ready for our first customers.

Right now we’re in a small office in downtown Palo Alto. But we plan on growing fast so things will change quickly. Uncertainty abounds: “...constant danger, safe return doubtful, honor and recognition in case of success.” (You'll be well compensated with a competitive salary and equity-based upside, comprehensive health, dental and vision coverage. But yeah... honor and recognition, too.)

We have big ideas for where Mashgin is headed and we would love for you to be involved. If you’re interested in joining the team email us at: jobs@mashgin.com

Mashgin Inc. (YC W15) | Computer vision engineer | Palo Alto, CA | Full-time

At Mashgin, we apply advanced computer vision to real world problems, starting with retail checkout. Our kiosk can scan multiple items at once without barcodes or RFID, reducing checkout time by 10x. But that's just the beginning -- we’re building the future of machine vision for all businesses.

We're looking for someone with serious engineering skills that can help us fulfill the promise of computer vision in the real world. You should have experience writing vision algorithms in C++. Bonus points if you've worked with 3D cameras, multi-cam scene reconstruction, or machine learning. As the first non-founding engineer you’ll get the chance to shape the foundation of Mashgin.

We're a team of 3 with previous experience in machine learning at Facebook and Microsoft, and computer vision for humanoid robotics at Bell Labs and Toyota. Mashgin is well funded from top investors following YC W15, and we're working hard on getting our machines production ready for our first customers.

Right now we're in a small, scrappy office in downtown Palo Alto. But we plan on growing fast so things will change quickly. Uncertainty abounds: "...constant danger, safe return doubtful, honor and recognition in case of success." -- Ernest Shackleton

If you're interested in joining the team email us at: jobs@mashgin.com

I've found a good rule of thumb in business is that you can only be successful in the long run if your value proposition is a win-win-win for you, your suppliers, and your customers.

When a majority of your customers are being unknowingly screwed over as you reap a huge amount of unnecessary producer surplus, it's a losing proposition in the end. Companies like Herbalife can just do it on a scale where it takes a long time to fizzle out.

Unrealized losses over a short time period have nothing to do with being "right" or "wrong" about a stock investment.

Fluctuations of investor expectations (i.e. stock price movements) have especially nothing to do with this article's analysis. It presents data and comes to a pretty reasonable conclusion from it. Whether you wan't to call it a "pyramid scheme" or not is semantics, the conclusion speaks for itself.

Short-term investors can make money betting in the opposite direction of long-term investors who will also eventually make money. Herbalife can be a pyramid scheme and short-term investors can make money as the stock price rises -- the 2 aren't mutually exclusive.

Any recommendations for overcoming this issue?

I'm working on a project with the eventual goal of creating a crowdsourced, interactive version of a map just like this.[1] We haven't gotten to this point yet (only to the 1930s at the moment), but we will eventually come across this problem. Currently we have "admin" levels (sovereignty > state/republic/colony > region/county > city) but this may be disrupted the further back in time we go.

One advantage of having data like this publicly crowdsourced is that a decision on how the information should best be processed and presented can be made by the community and potentially changed over time. (And now my pitch: help us out! We just started building the community but have big plans.[2])

[1] http://atlastory.com/

[2] http://forum.atlastory.com/

Atlastory - Palo Alto, CA

Atlastory is a bootstrapped startup working on a project similar to OpenStreetMap or Wikipedia, with the goal of mapping the world throughout history (and seeing maps change over time). Getting to this point requires building a crowdsourcing community around historic map data.

We're looking for other talented hackers to work with, preferably with experience in Javascript/Node. (Current stack is EC2, Postgres, Node, Rendr.js/Backbone.js.) More importantly others who are interested in history/mapping and in solving challenging problems.

email: max at atlastory dot com

Palo Alto, CA

Seeking fellow developer to help build the Google Maps of history

I’m working on a project, similar to OpenStreetMap or Wikipedia, with the goal of mapping the world throughout history (and seeing maps change over time). Getting to this point requires building a crowdsourcing community around historic map data.

I’m looking for other talented hackers to work with, preferably with experience in Javascript/Node. More importantly others who are interested in history/mapping and in solving challenging problems.

email: max at atlastory dot com

How about a PDF? :)

http://stuff.maxolson.com/On-Startups.pdf

There are other startup-related essays thrown in there too. PG, if you're reading this and want it taken down, please let me know and I'll oblige.

You "felt like a sucker" and from another comment "feel like I've been lied to"?

Do other people's comments and criticisms make you change your mind about something that easily? If you thought it was a good book when you read it, why would anything anyone else says change that? (Short of, say, Gladwell made up the entire thing or has been rampantly plagiarizing without attribution.)

Gladwell's books are very well-written narrative combining summarized research studies, anecdotal stories and histories to support his overall thesis. It is not a scientific study. He "connects the dots" of what he believes to be the central theme of his book. I haven't seen anywhere that he "lied".

There will always be detractors to any non-fiction author, especially one so popular. (See Nassim Taleb's books as another example.) If you can find a non-fiction book without major detractors, it probably isn't very good.

Also coming from Salt Lake City -- flying into SJ on Friday, and back Monday morning. Probably planning on renting a car, if anyone is interested in sharing.

I have thought of this but haven't really pursued anything yet. I also have permission from Buffett to publish his early partnership/hedge fund letters from the '60s, so will probably be doing that in the next year.

Most of the letters are freely available on Berkshire's website. The "extra material" includes a few early letters not on the web (got these from Buffett so the collection would be complete), and charts/tables/indexes that I put together myself.

The income is not shared with Buffett, but any income I receive is "return on capital" from me taking inventory risk. So I do have to invest some of my $$$ to buy the book inventory, and as it's sold to Amazon I make (great) return on that money. But the profit per book is actually pretty low compared to other published books -- I could be charging much more for the book, but it was part of our "agreement" that I make the total price of the book as cheap as possible. (Which I think is accomplished with Amazon's price of $19-25 for a 700 page textbook-size book.)

~$1000 a month. Designed, edited and published a book of Warren Buffett's letters to shareholders (see http://amzn.com/1595910778). An opportunity I stumbled into. But I suppose "get publication rights from Warren Buffett" isn't a repeatable business model :)

Self-published a book in April, which is now selling in Paperback and Kindle versions (http://www.amazon.com/Berkshire-Hathaway-Letters-Shareholder...). It's been doing pretty well although there was a built-in audience and it's not too difficult to market a book when the author is Warren Buffett.

I did a printing of the book and am selling the physical version on Amazon Advantage. Advantage takes a 55% cut of the List Price, and handles all shipping, ordering, etc. I just have to send them inventory when needed.

The program works, but so far in my experience it's a huge pain in the ass. They send out Purchase Orders when they need more inventory and it's pretty unpredictable (an order could be 4 copies or 900, which means shipping really changes my COGS). They also "lost" a shipment of 800 books that took them a month to find. So I wouldn't recommend Amazon Advantage unless there isn't another option.

Hacker News readers who are interested in how share buybacks work and how they can be a very good idea (even in the tech industry) should read up on Henry Singleton & Teledyne:

http://en.wikipedia.org/wiki/Henry_Earl_Singleton http://observer.com/2003/04/the-brain-behind-teledyne-a-grea...

Think of it as a way to reinvest in the future of your current business without having to actually expand it (which, in the case of a company at Apple's size/stage, would likely be wasted on low-returning incremental investments). Apple returning capital is a good thing -- it's how businesses should work.

Mistakes have certainly been made, but yes: this is typical Wall Street fare.

When Johnson first started the turnaround over a year ago, he made it clear this was a long-term undertaking, of at least 3 years, but probably 5 years. That should really hold true for any turnaround, but especially one of this size.

The company is only 1 year into the turnaround. It's headlines like this and investor myopia that gives way to great investment opportunities. Maintaining a long-term view can frequently lead to making money off of those who don't.

> Not discounting works for Apple because it's the only store selling MacBooks.

The only store... along with Best Buy, Amazon, Simply Mac, etc. etc.

I can think of a lot of scientific "big data" applications that would be very challenging problems. There are many domains that need to create & automate better systems that Collect > Organize > Parse > Visualize data. Weather, climate, ecology/species, traffic systems, etc. To do these right would be very challenging and would have a big impact on the world.

Now some self-promotion: I'm looking for a full-time partner to help me with Atlastory, an app that visualizes all of history on a map. This fits with the idea above. It's a simple concept, but there's huge engineering & design challenges involved: imagine all the geographic data involved in Google Maps, now multiply that by 1,000, and figure out the easiest way to obtain/retrieve/display that data.

The project I'm currently working on is called Atlastory, and would be a Google Maps-like interface throughout history. The idea is to eventually have political maps like those in the OP + specific maps for historic events.

Demo is up at http://atlastory.com (for now only maps are WWII and Civil War). If anyone's interested in helping out please contact me. (Currently looking for another developer to help with the much more complicated back-end)

You're right -- should have thought about that. I removed any specific information obtained from FundersClub from the post. The rest of the analysis is from other public sources (now listed at the bottom).

Atlastory, Inc. | Salt Lake City, UT | Full time

About: Atlastory visualizes history on a map, allowing users to explore the world through space and time. It’s like Google Maps with the ability to look 10, 50, or 1,000 years in the past. This is an application that has the potential to change the way people around the world learn history and interact with the past. This is a seed-stage startup with funding.

Job: this is an employee #1 position for a generalist back-end developer. You will be involved in developing the back-end map rendering, data organization, and system buildout. We use Ruby, Python, Node.js. You should have experience with DevOps/AWS products, database design (SQL, Postgres, etc.), HTML5/JS/front-end a plus. Most of all this set of problems requires someone who can learn as they go & figure things out that have never been done before. A few of the problems you'll be working on:

* PROBLEM: How to design a git-like system for collaboratively editing data that spans the history of the world.

* PROBLEM: How to organize a huge amount of GIS data so that it can be easily accessed, fed into an image renderer, backed-up, etc.

More details in the full posting here: https://docs.google.com/document/d/1bbbWWp7oPHcwbrqKlgJrp7i1...

This made me think of how awesome it would be to be teleported to a random place in the world like this, just walking on the street. You wouldn't know where at first, and you'd have 24 hours before being transported to a new location. <Any script writers out there?>

This explains why Groupon & LivingSocial have roughly maintained their market shares:

http://go.bloomberg.com/tech-blog/files/2012/08/livingsocial...

I think the primary reason for their "downfall" was not lack of barriers to entry (there were none at first, but once industry is established, there are as you mentioned), but a much smaller market overall. i.e., the "group coupon" model was very popular for a short time, much more so than its long term steady-state usage.

Atlastory, Inc. | Salt Lake City, UT | Full time

About: Atlastory visualizes history on a map, allowing users to explore the world through space and time. It’s like Google Maps with the ability to look 10, 50, or 1,000 years in the past. This is an application that has the potential to change the way people around the world learn history and interact with the past. This is a seed-stage startup with funding.

Job: this is an employee #1 position for a generalist back-end developer. You will be involved in developing the back-end map rendering, data organization, and system buildout. We use Ruby, Python, Node.js. You should have experience with DevOps/AWS products, database design (SQL, Postgres, etc.), HTML5/JS/front-end a plus. Most of all this set of problems requires someone who can learn as they go & figure things out that haven’t been done before. A few of the problems you'll be working on:

* PROBLEM: How to design a git-like system for collaboratively editing data that spans the history of the world.

* PROBLEM: How to organize a huge amount of GIS data so that it can be easily accessed, fed into an image renderer, backed-up, etc.

More details in the full posting here: https://docs.google.com/document/d/1bbbWWp7oPHcwbrqKlgJrp7i1...

Atlastory, Inc. | Salt Lake City, UT | Full time

About: Atlastory visualizes history on a map, allowing users to explore the world through space and time. It’s like Google Maps with the ability to look 10, 50, or 1,000 years in the past. This is an application that has the potential to change the way people around the world learn history and interact with the past. This is a seed-stage startup with funding.

Job: this is an employee #1 position for a generalist back-end developer. You will be involved in developing most of the back-end to enable map rendering, data organization, and system buildout. You should be well versed in Ruby, Python, DevOps/AWS products, database design (SQL, MapReduce/Hadoop, etc.), HTML5/JS/front-end a plus. Most of all this set of problems requires someone who can learn as they go & figure things out that haven’t been done before. A few of the problems you'll be working on:

* PROBLEM: How to design a git-like system for collaboratively editing data that spans the history of the world.

* PROBLEM: How to organize a huge amount of GIS data so that it can be easily accessed, fed into an image renderer, backed-up, etc.

More details in the full posting here: https://docs.google.com/document/d/1bbbWWp7oPHcwbrqKlgJrp7i1...

To further distinguish from an earlier HN post that also mentioned adjusted medal rank.

Ah sorry had to take it down, didn't know about the terms attached.

I made the PDF personally to turn into a printed/binded copy (I like physical books) with nice typography and after seeing the HN post thought I'd share.

If anyone still really wants this fancier version follow/msg me on twitter: @maxolson