Hacker Newsnew | past | comments | ask | show | jobs | submit | hurrycane's commentslogin

DatologyAI | Member of Technical Staff | Full-time | Onsite, Redwood City, California | https://www.datologyai.com/

At DatologyAI, we build tools to improve large-scale deep-learning models by selecting the right training data. We are automating data curation at petabyte scale, helping customers to train models faster, to better performance, and enabling smaller models to achieve performance competitive with much larger models. Models are what they eat, and the data models ingest determines everything about their capabilities.

Since founding less than a year ago, we have raised $57.6M, including a $46M Series A in May 2024. Our backers include AI luminaries Geoff Hinton, Yann LeCun, Jeff Dean, and Amazon and Microsoft.

We are hiring: Research Scientist: https://jobs.ashbyhq.com/DatologyAI/d5090b17-8b58-42ee-afa6-...

Data Infra Engineer: https://jobs.ashbyhq.com/DatologyAI/7a64144a-c612-4af8-bdf8-...

ML Engineer: https://jobs.ashbyhq.com/DatologyAI/c64f9ea8-0943-476d-aabd-...

ML Infra Engineer: https://jobs.ashbyhq.com/DatologyAI/b5a42904-d7ec-45a4-96df-...

Full Stack Engineer: https://jobs.ashbyhq.com/DatologyAI/5546e184-6f86-4540-868d-...

See more details on our mission here: https://www.datologyai.com/post/introducing-datologyai-makin...


Here's a blog article regarding the changes required for search to support the tweet length changes: https://blog.twitter.com/engineering/en_us/topics/infrastruc...


Unfortunately behind a pay-wall.

Full text:

Fitbit, the leader in the fitness band market, is near a deal to acquire smartwatch maker Pebble, according to three people briefed on the deal.

The price couldn’t be learned but it is thought to be for a small amount. Pebble had been looking to sell, one of the people said. There have been signs over the past year or so that Pebble was facing financial challenges. Earlier this year it reportedly laid off about a quarter of its workforce.

THE TAKEAWAY The expected sale of Pebble to Fitbit signals a consolidation in the wearables market.

The Pebble brand will be phased out after the deal. What Fitbit will get is Pebble's intellectual property, such as its operating system, one of the people said.

The deal signals a consolidation in the wearables space, which has been crowded with several players including Apple and Jawbone. Pebble burst onto the scene several years ago with a flashy Kickstarter campaign that drew interest from consumers. But it struggled to gain traction.

Pebble burst onto the scene several years ago with a flashy kickstarter campaign that drew interest from consumers. Fitbit, which went public last year, has had its own issues. Its stock price which has fallen from a peak of $49 in August of last year to roughly $8 today, giving it a market cap of $1.87 billion. The company recently reported net income fell 61% to $43.5 million in the nine months to Oct. 1, despite a 39% increase in revenue to $1.6 billion. Fitbit's main product is fitness bands but it sells what some have called a smartwatch, pitting it directly against Apple's watch.

Meanwhile, Jawbone, another maker of wearable devices, has also struggled in the past year or two and had looked for a buyer, without success.

Source: http://go.theinformation.com/dfNa4p4B_IY


Somehow this whole startup thing often puzzles me.

I read good things about RethinkDB, bad things about MongoDB and almost nothing about ArangoDB. But Rethink closes shop, Mongo is used everywhere and Arango gets big funding.

Same with Pebble, read only good things about them, now they sell for a "small amount".


Last year there were 233,000 petitions. Only 85k petitions are actually accepted each year. Compare that to 172,500 petitions received in 2014 and 125,000 in 2013.

This means the rate of companies that use h-1b has actually slowed down by a lot. I know as a fact that most of the US tech companies that have offices in Europe, are actually transitioning to using L-1 visas and don't apply for H-1b anymore.

If you're a small startup and you want to hire someone who is not a US citizen your options are pretty limited right now.


So true! Been there myself! And have friends that are getting told by big tech companies that they don't do H1Bs anymore and prefer to do L1s and people working from their Europe offices.



Why is that? Finagle, Mesos and Aurora look like solid open source projects from Twitter.


Aurora is from Twitter, but Mesos is not. They hired some of the team that developed Mesos while in academia, but still. Mesos is external of Twitter.

Disclaimer: I use both Mesos and Aurora.


I started looking at finagle, finatra and scrooge. i love all three projects. but at the same time, the documentation is really really bad.

if you have the same requirement as twitter, nice -- but if you detour even a little bit, you are going to have a shitty time (for example, for thrift, i wanted to use buffered codec instead of framed and there was not a single document explaining how to do it. i spent ~2h perusing unit tests to find a way which i don't know if it's correct or not).


Also forgot to mention scalding.


Don't forget Netty..


netty is not a twitter project. At various times the major committers have been employed by JBoss/RedHat, Apple, etc. etc.


I'm curios what tech stack they use for this?


No idea what Pinterest are using, but I led a team building the same thing using (mostly) commodity search kit in 2008.

Feature extraction was done with standard Java libs (proprietary algorithms though). Queries were initially performed using a vector space model, but I moved that to using an inverted index (Lucene) because in our use case the image queries were usually combined with free text and parametric search params.

The main issue we faced was scaling search with large number of query parameters, since a naive implementation created something like 300 query terms for each visual search. We did various things to optimise that, from distributing the index, to using index statistics to pick optimum words to query. I submitted some optimisation code (a modified MoreLikeThisQuery with an LRU term cache) back to Lucene, not sure what happened to it, think the JIRA issue is still open.


Could you give a high level description of the features used? Interest points, corners, edges, color histograms, or something more sophisticated?


The context was fashion retial, so we used interest points & edges to match similar shapes, and histograms for colour similarity.


Google Image search has such a feature too. One can drap&drop a local image to the search bar on the Google Image page. Help page: https://support.google.com/websearch/answer/1325808?hl=en ; info: http://www.quora.com/What-is-the-algorithm-used-by-Google-Se...

One of the earliest such similarity image search was a research prototype on Airliner.net (ca. 2006): http://www.airliners.net/similarity/ , e.g. http://www.airliners.net/search/similarity_search.php?photo_... , http://infolab.stanford.edu/~wangz/project/imsearch/SIMPLIci... , http://alipr.com/cgi-bin/zwang/regionsearch_show.cgi

It's called Reverse image search: http://en.wikipedia.org/wiki/Reverse_image_search


Here are some of the technologies we use:

Caffe - deep learning feature computation and model training

OpenCV - local + other features

Zookeeper - service discovery

Cascading - batch processing jobs

We’ve built infrastructure around some of these libraries to operate at scale. For example we’ve build an incremental feature extraction pipeline that at the core uses caffe + opencv for the feature extraction (more details on how this works is in the paper).

We serve our infrastructure through ec2.


Each year there are 20k work visas for people who graduated MSc in the US and 65k work visas for the rest of world.

That means if you got an offer to work from a startup or a big tech firm there are about 23% changes you'll get the visa.


I had the same issues last around July - August and continued upgrading as they release new versions and somewhere along the way it got fixed.

You can also had to fine tune some timeouts (election and compression if I remember correctly) to get the best performance out it.


I strongly believe that this is the next version of Phusion Passenger.


The writing style is very similar.


Their architecture looks to me really mature - a lof of features are the same as Passengers'


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: