HNHacker News
TopNewBestAskShowJobs

tixocloud

935 karma · joined March 7, 2012

Founder/CEO, FinBright - Financing platform for sustainability initiatives (http://finbrighthq.com). Email me at teren at finbrighthq.com
submissionscomments
tixocloud··on Ask HN: Who is hiring? (June 2019)
Orchestra | Front-end & Back-end engineers | London/Toronto | On-site | Full-time

Orchestra helps data scientists deploy machine learning models to production faster whether it's building a Docker image, serving as an API, creating batch processes or integrating directly to enterprise systems. Effectively, we are DevOps for ML and we automate deployment and release management of machine learning models.

We're still an early stage startup looking to close our seed round in the next few months. As an early member, you'll have the opportunity to define the space and set the standard for what it means to do DevOps properly within ML. The platform is currently developed using AngularJS, Python, Docker/Kubernetes.

Drop me a line at teren@orchestrahq.com if you love ML and want to help get machine learning models deployed faster, at scale and in a variety of shape and form.

tixocloud··on An End-to-End AutoML Solution for Tabular Data at KaggleDays
Auto ML is great as long as there is some rational on why a particular model is developed. However, in real world applications, there are so many constraints and considerations needed besides just having the most accurate model.

Furthermore, as a VP of data science, selling the business on the value and benefits of data science is non-trivial as many are not even aware of what's capable with data science so my personal opinion is that ML isn't quite suitable for mainstream work. Practitioners still need to have a strong grasp of the business context, data features and nuances, etc. which still is quite technical in nature. Asking a data analyst to build models with auto ML is something I consider to be a leap too far and is risky to put into production. The only way it could work is to have an experienced data scientist supervise.

tixocloud··on Local-first software: You own your data, in spite of the cloud
Local computation makes sense especially from a user experience perspective. However, another challenge arises from talent availability and willingness to work with this sort of deployment model. More complex deployments generally lead to the need of hiring really top notch engineers, which as we know is quite scarce. The next best thing might be infrastructure that makes the deployment easier but no one's really interested in investing in infrastructure tech without some tangible reward.
tixocloud··on Ask HN: Are you put off building something because it already exists?
Having been through the startup journey for the past 4 years, this advice is solid especially around the passion.

Having passion for your idea is quite understated. Passion is what pulls you through the tough times and helps truly unlock the magic in your solution. It's because you care about solving the problem that gives you the competitive advantage.

In my early days, I only spent about 3-6 months on an idea trying to validate it, hitting a roadblock and feeling deflated. With my current product, the idea has been worked on, thought about, validated and iterated for close to 6 months. Long time in the startup world but it's important for me to truly understand what the market is looking for before diving straight into sales. This turned out to be quite important as potential enterprise customers to make sure our solution resonates.

We're now looking to land demos and hopefully close some early sales and then look to raise seed funding once things have been derisked.

tixocloud··on Ask HN: What does your ML pipeline look like?
We deal with fairly large volumes of data on a frequent basis so it would not make sense for each data scientist to create a copy within their own environment. Everyone works off a centralized data source and we provide them with Jupyter/Spark in an internal cloud environment.
tixocloud··on Ask HN: What does your ML pipeline look like?
We use Jenkins for our builds (and soon simple testcases), which then feeds into our system that's built from scratch. However, we are looking to productized the system as we had some discussions with other corporate partners who are interested in as well.

The system we've built in-house is fairly simple to keep development fast - versioning is manual (https://packaging.python.org/guides/hosting-your-own-index/) but no reason why you couldn't use a repository manager.

For monitoring, we essentially track any activity related to the model including inputs, outputs, timestamps, duration, etc. to a database and have JavaScript charts render. We might put this into Kafka but seems overkill at the moment and likely force us to hire an actual support team.

tixocloud··on Dear AI startups: Your ML models are dying quietly
Your startup looks quite interesting and I can actually see potential for commercial usage. How's the growth rate been?
tixocloud··on Ask HN: What does your ML pipeline look like?
+2 for bringing the (ML/stats) code to the data instead of the other way around
tixocloud··on Ask HN: What does your ML pipeline look like?
We're building a deployment system based on the following principles that we've learned:

- Version control of models and datasets

- Code, naming conventions and formatting consistency

- Testing of models before deployment into production (in most cases, this helps gain credibility across engineering)

- Model review process with a senior data scientist

- Kubernetes/Docker for deployment serving as an API or running a scheduled job

- Model performance monitoring when in production to identify degradation

tixocloud··on Ask HN: What does your ML pipeline look like?
We're in the financial services space. The data science team has access to a data warehouse that gets updated daily but no access to production data directly so we use this for development and experimentation.

The definition of production environment varies for us. Production could mean running scheduled batch jobs on a separate 'production' pipeline from the data warehouse and results can stored in a database. Naturally, all our data is in-house at the moment so we have had to set everything up ourselves.

We're working on an enterprise-wide deployment system that integrates with the build/testing infrastructure where all the data science teams across the financial institution can leverage for deploying models and monitoring performance.

tixocloud··on Ask HN: How do I improve our data infrastructure?
> Everyone else is a consultant, and they don’t have any incentive to care about the long-term impact of architecture choices: their management only evaluates delivery (graphs, model metrics, etc.). How do I go about raising awareness?

It sounds like management treats data science as a service function as opposed to a strategic function. Do you get the sense whether they do have any incentive to grow the data science team properly as opposed to leveraging consulting resources? My gut is that there are trending to building out a data science team given you're their first data science hire but it also means that your results will largely dictate what the team will evolve to in the future.

To raise awareness, get a good understanding of the management team and their motivations. Frame the issue to something that they are most concerned with (likely business-related) and highlight the impact of not addressing the issue of poor architecture.

Present your findings, highlight the impact and go a step further to propose 3 options. Highlight why the option you're recommending is the best.

tixocloud··on Ask HN: One-person SaaS apps that are profitable?
Really great idea! How do you measure the softer criteria like “Eats lunch together”?
tixocloud··on Ask HN: Machine learning resources for audio processing
Here's a resource that breaks down the various audio processing tasks and provides case studies: https://www.analyticsvidhya.com/blog/2018/01/10-audio-proces...

It's slightly academic so here's a more practical resource: https://towardsdatascience.com/audio-classification-using-fa...

tixocloud··on How do data scientists log their workflow/process?
Given that we're in a very large organization, Excel spreadsheets have been our default from the start. I created a template to help our data scientists document issues (we also use Confluence but we haven't got it in a good state just yet).

For actual tracking, we use rough estimates - more so because we want to balance being a pain of managing the data vs the value that we get out of it.

tixocloud··on Google launches an end-to-end AI platform
Agreed with the long term benefits on fitting into existing architecture.

I did speak with someone in financial services who’s been all GCP and they are quite impressed by the way everything is integrated and how they do not have to shift data from storage to train.

tixocloud··on Google launches an end-to-end AI platform
There’s plenty of room to compete with Google and it makes things easier when it comes to market education.

We’re also a startup in the ML space but solely focused on production deployment.

You’re right in that model accuracy is not the most important thing. There are many other considerations as to why a model should be used including training time, processing costs, value.

Furthermore, as someone mentioned earlier, Google’s service is horrendous and provides another angle to address. Their reputation as a company to be trusted with data is also somewhat shaky given all the privacy concerns.

tixocloud··on Google launches an end-to-end AI platform
Are you guys quite embedded into the Google Cloud system?

We’re a startup that helps with AI deployment management (our only focus is production AI - embedded everywhere) and I wonder how many companies are who’d be interested in their end-to-end offering.

tixocloud··on Google launches an end-to-end AI platform
Which industry are you in? Would love to get your thoughts on the IBM platform.
tixocloud··on Ask HN: Launching a Product on Show HN?
If it's not for developers, then you'd miss the valuable learnings from your target market. I'd advise you to find a more suitable place where your target market is.
tixocloud··on Ask HN: Debitize is shutting down, any alternatives?
I’ve worked in credit cards and this business model (like most fintechs) isn’t sustainable. There’s just a fraction to be earned on transaction fees so not surprised it’s being bought out. I'd say for many alternatives, they will eventually follow the same route.
tixocloud··on A16Z is re-registering as a financial advisor, renouncing its status as a VC
Did you guys ever eventually work with VC firms who didn't understand your product? I get a sense that there might be a big mismatch but wonder how much that really affects things day to day.
tixocloud··on A16Z is re-registering as a financial advisor, renouncing its status as a VC
Great note to hear that there are legitimate VCs like a16z who aren't just after their own interests. Did you work with other investors as well? If so, would love to contact you privately regarding your experiences as I'm looking to fundraise soon.
tixocloud··on Ask HN: Machine learning contractors, how much do you charge hourly?
What kind of work are you looking for?
tixocloud··on Ask HN: Machine learning contractors, how much do you charge hourly?
Interesting regarding the 70% - do they actually you bring you in as a consultant to help them map out their ML strategy?
tixocloud··on Ask HN: Are there any healthcare software SMEs?
Funny - looks like everyone is moving towards the same model (i.e. marketplace and ecosystem) which is a logical step! Wonder what would happen if there are too many marketplaces.

Effectively, I won't be recommending treatments but rather I'll be hosting ML models from other researchers that might - I wonder if I can write a legal clause that clears me from FDA approval.

It seems like IBM has a way of saying 'not FDA-cleared for diagnostic use': http://www.merge.com/Solutions/Cardiology/Merge-Cardio.aspx

tixocloud··on Ask HN: Anyone familiar with medical software regulations and certifications?
Thanks for the overview - that's definitely very helpful.

From my perspective, it looks like I'll have to deal with both as we're looking to host medical imaging algorithms and even if we aren't the ones providing the recommendation itself, we will still be helping research labs process them.

Where can I learn more about FDA requirements and HIPAA certifications? Are there organizations that eventually conduct an audit and provide confirmation that software companies indeed comply?

tixocloud··on Ask HN: How can we improve the U.S. recycling industry?
We can think about making existing products reusable or creating new products from previous materials.

There are a few furniture manufacturers who are taking plastic from the ocean and recreating them into high-end designer furniture.

When creating new products, the functionalities need to closely replicate what plastic has been able to do. There's not enough net benefit to convince people to switch to alternatives at the moment.

tixocloud··on Ask HN: Document Management for Startups
What type of documents/paperwork are you looking to archive or extract? I would evaluate whether you need to make everything searchable or prioritize certain documents over others.

For example, we're working on a process to scan through our emails, pick out invoices, archive the document but extract the necessary information for accounting purposes.

For other documents, we'd like to make them searchable but they just get dumped into a cloud storage space at the moment.

tixocloud··on Ask HN: Part-Time Work for Founders?
It certainly felt that way. Did design firms eventually ask whether they could showcase their work on your site?
tixocloud··on Ask HN: How to Get Good at Sales?
Like most things, getting good at sales is about practicing, learning and practicing again.

Books and advice are great to get started but eventually there is no substitute to actually doing it yourself.

That being said, we are all actually “selling” in our day to day lives without formally defining it as sales. From convincing your parents to buy you a toy all the way to landing the first job involves similar techniques.

For a really simple sales weekend task, try buying a pack of bottled water and see how many you sell.

← PreviousPage 7 of 34Next →