jasonmar

better be prepared with an AI tool to monitor all sales calls and forcibly terminate the zoom immediately if anyone mentions assembly

A single neuron has hundreds to hundreds of thousands of synapses and at each synapse there is a number of receptors and channels on the receptor end and neurotransmitter molecules and vesicles on the transmitter end. Synapses are far from the nucleus so there are reservoirs of mRNA hanging out waiting to be transcribed that needed to be staged there. There's epigenetic state affecting how much and what type of mRNA is produced. All of these are continuously affected by the combination inputs. How much of this needs to be included in the model? I'd like to see them comment on which parts of the original system they consider out of scope.

Does Sourcegraph support Scala, COBOL, RPG, PL/1? Beyond the languages in the docs, how much work is required to use any of these languages? https://microsoft.github.io/language-server-protocol/impleme...

What does the client need to do? Which would be more useful, desktop GUI or command line? Would it be ok if the client runs on the jvm? Should it be deployed as a systemd service?

It would be fantastic if Moderna can manufacture and deliver the mRNA into human cells cost effectively and at scale.

Outside the cell, there are enzymes that cut up mRNA. So lipid nanoparticles are used to protect the mRNA. But then even if this makes it into the cell, the nanoparticle itself can be ejected by cell machinery.

If the mRNA can avoid ribonuclease long enough for the mRNA to be translated into protein rather than being degraded, then it seems pretty straightforward that the protein may be bound by MHC and make it to antigen presentation, T cell activation, B cell activation and hopefully affinity maturation and Memory B cells for the viral protein.

This paper explains some of the challenges: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6453548/

As part of my job over the past 2 years I’ve been migrating a data warehouse with data being loaded from Mainframe into BigQuery. I’ve learned a lot and gained respect for the platform.

One of Google’s differentiating products is the TPU, to train neural networks in hardware. Meanwhile, the mainframe has specialized hardware for TLS, gzip compression and encrypted storage.

If it's a critical system and you don't have the ability to fix it but you know it's only a matter of time before it fails, the sensible thing to do is change jobs.

I joined Google less than 3 months before the walkout. Mid level management practically encouraged us to attend. There were big numbers but most were probably there out of curiosity. I was really disappointed. In the New York walkout, there was a tiny area with an underpowered bullhorn and nobody could hear what they were saying. Exactly one hour after it started, everyone was bored of standing around and went straight back to work.

The author of the article is a professor of marketing so I'd be interested to learn his opinion on WeWork's marketing instead of accounting, corporate structure and governance.

Adam's unconventional moves appear extremely outlandish when spelled out in the S-1 disclosures, but because I have no position I just find it entertaining. I'm looking forward to the conference calls.

I worked for WeWork for a short period of time during one of the SoftBank rounds, and one positive unconventional thing the company did was give employees with vested options the opportunity to cash out alongside Adam, rather than being forced to wait until post IPO lockup.

See this example for how neural network can be used to "sort and search by vague similarity and classify on a spectrum of multiple qualities". youtu.be/5PNnPagENxQ?t=1540

Descartes Labs uses the pre-trained ResNet 50 and removes the final layer which does classification. What's left is a layer that provides image features necessary to do classification. These features can be used to sort images by similarity and search for similar images.

Publicis Media | Scala, Python, C++, Ruby, Spark | New York, NY | ONSITE | Full-Time | http://www.publicisgroupe.com/en/services/services-publicis-...

Publicis is a top 4 global advertising conglomerate.

Publicis Media is investing heavily in building a new technology platform for managing all aspects of cross-channel and multi-device advertising campaign planning and management. Agency clients are top-tier global brands.

I'm on the data team and the working environment is great, our people are very experienced and easy to work with. The office location is ridiculously good.

We're looking for new team members with strong Computer Science fundamentals and Software Engineering skills to help us build data pipelines and data access services with good performance, high test coverage, and robust operational characteristics. The UX team and many others also have a lot of openings.

Check my profile for job links and my contact info.

Audible | Newark/Jersey City NJ | ONSITE

Build machine learning infrastructure at an Amazon subsidiary that still runs like a startup.

We are using Scala/Akka/Python/Spark to build tools and services for a data science team. There is a lot of opportunity to make an impact.

Our offices are 20-30 minutes from New York City.

Email is listed in my profile.

HN is always talking about metrics, scalability, and engineering, here's my take on this from that perspective:

Imagine a twitter engineer working on the redesign was looking at data from the homepage, looking for something to remove.

It turns out that the least clicked on thing on the entire page is the client attribution.

He looks at eye-tracking data too, and it turns out the only people who look at the client names, are developers!

The engineer opens his iPhone's calculator, and finds that by removing attribution he is reducing dynamic stream content by 33% (tweet, time, attribution) for 100M+ users.

What kind of engineer is not going to jump on this? Especially for a piece nobody except developers is going to complain about losing?

A cleaner, faster homepage is better for users. Sorry developers, the world does not revolve around you.

You are probably not going to be able to do enough volume initially to get a discount on your execution. In futures the top tier requires 10's of thousands of contracts per month, in equities you'll need to be trading millions of shares per month. It would be extremely hard to overcome this even with great algorithms.