News for hackers. Stuff that matters.

philth

Big fan of projects like BentoML, the whole ML space is powered by some truly great projects. What we've tried to do with truss is create an experience that is optimizing for simplicity and universality. Our roadmap begins to deviate away from projects like BentoML, and so creating truss was very much motivated by the need to continue this mindset as we march down through that roadmap.

context: I am one of the contributors

as the sibling comment mentioned, the other two founders were recently on a podcast with more details, but i can give the TLDR here: we've all worked as ML practitioners or supporting them in engineering roles. We realised there are common patterns in establishing a ML powered product, eg deployment, integrating into business logic/databases, building UI to prototype functionality and eventually turning that prototype into a functional application. We've built the blocks that allow you to put these together in novel ways without constantly reinventing the wheel and being distracted by common pitfalls.
Hi HN, I'm one of the founders and would love to give some context. Our aim is to make building machine learning powered applications as painless as possible. We built the product that we wish existed when we were flailing about, trying to stand up ML applications in the past.

We’ve built a product that allows users to seamlessly integrate ML models with their custom business logic, expose them as APIs, and ultimately assemble user-facing apps that utilize those APIs. Many ML practitioners find dealing with infra (e.g. docker, k8s) and front-end engineering to be daunting tasks. We want to show them that they can achieve a lot more (e.g. build full-fledged ML-enabled user-facing applications) without having to learn those skill sets.

Our aim was to build something that was 1) simple, 2) powerful, and 3) a joy to use.

This is our take on that — we’d love to hear what you think

Hi HN. We’re excited to launch BaseTen today. Our aim is to make building machine learning powered applications as painless as possible for data scientists and engineers.

We’ve built a product that allows users to seamlessly integrate ML models with their custom business logic, expose them as APIs, and ultimately assemble user-facing apps that utilize those APIs. Many ML practitioners find dealing with infra (e.g. docker, k8s) and front-end engineering to be daunting tasks. We want to show them that they can achieve a lot more (e.g. build full-fledged ML-enabled user-facing applications) without having to learn those skill sets.

Our aim was to build something that was 1) simple, 2) powerful, and 3) a joy to use.

This is our take on that — we’d love to hear what you think

Re:Ask HN: Who is hiring? (May 2017) on Monday May 01, 2017 @03:14PM (#14240546)
Starling | Front End Engineer & Full Stack Engineer & Security Engineer | SF | Full-Time | Onsite

Starling exists to make organizations better. We're an aggregation and analysis platform for People data, helping companies create data-backed strategies to build great organizations. From diversity and inclusion, to attrition risk and prevention, we cover a wide range of challenging problems.

We’re looking to hire Front End, Full Stack and Security engineers. You would be joining a small, fun team of incredibly gifted and passionate peers.

Our platform is live and revenue generating, backed by some great investors. We're working with some of the top tech and non-tech companies in the US.

We make extensive use of python/flask, React, Node.js, D3.js - experience with analytics products is a huge plus. As we're still small and growing rapidly, we can't accommodate recent bootcamp grads.

Help us shape the future of work. Reach out to us at [email protected] to apply. Competitive salary/equity + all the perks

Re:Ask HN: Who is hiring? (March 2017) on Wednesday March 01, 2017 @12:39PM (#13766087)
Starling | Front End Engineer & Full Stack Engineer & Security Engineer | SF | Full-Time | Onsite

Starling exists to make organizations better. We're a data aggregation and analysis platform for People data, helping companies create data-backed strategies to build great organizations. From diversity and inclusion, to attrition risk and prevention, we cover a wide range of challenging problems.

We’re looking to hire Front End, Full Stack and Security engineers. You would be joining a small, fun team of incredibly gifted and passionate peers.

Our platform is live and revenue generating, backed by some great investors. We're working with some of the top tech and non-tech companies in the US.

We make extensive use of python/flask, React, Node.js, D3.js - experience with analytics products is a huge plus.

As we're still small and growing rapidly, we can't accommodate recent bootcamp grads.

Help us shape the future of work. Reach out to us at [email protected] to apply. Competitive salary/equity + all the perks

Re:Deep Learning Without Poor Local Minima on Tuesday May 24, 2016 @05:20PM (#11765598)
There is a great post [1] on stack overflow that might help contextualise this development.

[1] http://stats.stackexchange.com/questions/203288/understandin...

Keep us posted on your discoveries. It would be interesting to see how different the embedding is to word2vec trained on a different corpus. I imagine borrowed words like "python" are clustered with programming languages rather than snakes in this case.

As a side note, not really having looked too deeply into word2vec, does word2vec capture multiple meanings? If so, how?