AMA with Dean Attali, R Shiny consultant

Dean Attali AMA

As part of our Code For Cash community, we invite guests to come into our Slack channel every once in a while for an AMA session. Recently, our guest was Dean Attali.

Dean Attali

Dean Attali is the Founder & President of AttaliTech Ltd, a software consulting company. He currently specializes in R, but he’s worked extensively in Silicon Valley. Prior to founding his consulting firm, Dean gained experience in both large companies (such as IBM and Google), and small startups including and So, he’s done extremely well for himself in the “startup lottery” – and I know you might dismiss it as luck, but those of us who know better understand that luck is preparedness meets opportunity. Join us for the AMA to pick his brain and benefit from proximity to his success and influence.

The AMA!

The transcript...for search engines :)

zackburt [2:06 PM]
@channel Dean Attali (@daattali) is going to be doing an AMA over the next half hour or so about his experience as an R consultant.  Welcome, Dean!

@ali you might be interested in this one

Dean, how did you get started with R?  Why R? How did you find your first clients?  And what kind of work are you paid to do?

daattali [2:08 PM]
hi everyone

my background is in computer science, so like some of you (most of you?) I also used to mock R..... when you work in silicon valley you don't really hear about R much unless you're specifically in analytics after all

after a few years of work I wanted to get a taste of academia so I went for a masters in bioinformatics, and in academia and especially biology R is very strong. So everyone around me was using it.

zackburt [2:11 PM]
@tamachado thinking of you here

daattali [2:13 PM]
As a complete coincidence, the first course I took in my masters was how to explore data in R, and it was taught by a woman who's one of the most known figures in the R world, so I was lucky to be exposed to R by someone like her.  After working with it for a few months I started to really like it for data analysis, it was fun. And since then I've just been using it almost exclusively. It also helps that most R users and people who wrote R were not computer scientists - they're mostly statisticians, biologists, ecologists, etc. So coming in with a CS background was a huge advantage for me and I very fast became sort of an expert

I never tried to find clients, my whole path into consulting has been accidental. I used to write very popular blog posts about R, I had lots of popular answers on stackoverflow, I made a lot of contributions to R packages, I wrote some popular R packages, and so my name organically became known in the community. I never meant for that, I was just having fun with R. Eventually people started emailing me asking for paid help. At first I was scared and I was taking $50/hour thinking that's a lot (I was a student!) but as time went on I decided to just call myself a consultant officially and started charging more and more because there's too much demand

zackburt [2:16 PM]
So we’re to understand that you find 100% of your clients organically.  You have never done any outreach to find a client, people just find you?

Entirely through ‘inbound marketing'

daattali [2:17 PM]
People usually ask me for help in various things: audit their code, optimize their code, build a package (library) for them, build a web application in R, teach them how to use it

zackburt [2:17 PM]
You can build webapps in R?!

daattali [2:17 PM]
I tried AdWords for 2 weeks but I stopped it because I was getting over a dozen emails every morning and it was too much to handle

yes that's my specialty area, there's a framework in R called Shiny that lets you build interactive apps with R code

zackburt [2:18 PM]
So you were doing AdWords and people were clicking on your ads?

What was the landing page you used?  What were the keywords you used?  What was the text of the ads?

daattali [2:19 PM]
it's a fairly niche area so I think the conversion rate is fairly high

zackburt [2:19 PM]
Would you consider re-activating that and then delegating the excess work to members of our community?

I would love as many details as possible here

daattali [2:20 PM]
I haven't done ANY ad optimization, I really know nothing about advertising. So I'm sure it was terrible. My landing page was simply my "portfolio" page which is really quite terrible, but for now it works fine. My keywords were simple "r developer" "r consultant" "r expert" etc. The text was something similarly generic and crappy - but again, it's fairly niche, and my name is well known to R users, so I think it worked fine

I did start subcontracting, but I'm very particular about who I use. I've rigorously interviewed a few people who replied to a call of mine for R programmers

i assume (perhaps wrongly?) that most people here don't use R

dain [2:22 PM]
can you describe what your last client project was like, at a high level?

daattali [2:24 PM]
I have several projects going on at every moment, so I don't have a "last client" :slightly_smiling_face: The last thing I did for a client, yesterday, was to a data analytics company in Australia who built an app in R but it was very slow - taking over 30 seconds to load up. They wanted me to look at their code and find ways to make it faster.

ali [2:26 PM]
@daattali Any interest in bringing on new subcontractors who are R developers / data scientists?

daattali [2:26 PM]
It mainly has to do with having to know the technical details of R so that you know how/why some functions are slow and how to get around that. Over the weekend I helped a client with a very fun project: they had an application that lets a user manipulate a plot with different parameters, and the plot updates automatically. What they wanted help with is making that reproducible: so that the underlying code that updates the plot would be automatically available to the user

I'm always open to new subcontractors, but you would need to really understand R at a deep level, and be an expert in either shiny or visualizations (static and dynamic). If you are very comfortable with hadley's advanced R book, then that's a good first step

understanding NSE is very important. Understanding why people often say you should avoid loops, and when it doesn't matter. Knowing the ins and outs of S3 and S4 objects

dain [2:32 PM]
thanks for the answer!

zackburt [2:33 PM]
I apologize if this is too personal but how much do you charge now?

Trying to understand if it’s a good investment for budding data consultants to specialize in R

Also, how many years were you working in R before you started feeling confident enough to consult?  I understand that this is factoring in that you had a software & CS background

ali [2:35 PM]
Rocking AMA!

daattali [2:36 PM]
I charge per project usually, not hourly

zackburt [2:36 PM]

daattali [2:36 PM]
i find hourly is too awkward

which i know is not the typical way to go about it, but I also have no idea what I'm doing :stuck_out_tongue:

zackburt [2:36 PM]
That makes sense, you can parcel out some of the project to a subcontractor and still work congruently

No I think you’re on to something

daattali [2:38 PM]
i charge assuming i'll do all the work. I still look at all the code my subcontractor writes and have feedback to them. I only pass on tasks to subcontractors, not whole projects. But that will likely change in the future, I've only been doing this for a few months

hyzypg [2:38 PM]
Do you believe there's enough demand for data consultants to get R only/dominant projects? Data analysis seems to still be dominated by Python, but I'm new to all of this :slightly_smiling_face: (edited)

daattali [2:39 PM]
R vs python is a huge language war.... it's hard for me to know exactly because I'm very deeply involved in the R world and all i ever see is people using R. I would think python is still bigger

I wouldn't go into R thinking there's an untapped market. Also, R is very dominant in academia which won't pay you a lot for your projects

hyzypg [2:40 PM]
Ah, gotcha. Makes sense!

Do you ever come across clients requiring work to be done in both R and Python?

(not sure if that would be redundant?)

daattali [2:40 PM]
but it's growing. It's mostly small companies/startups that come to me, they want to do analytics in R, but I would still say python has more usage

I do have clients sometimes asking me to do things that I tell them R is simply notthe best tool for that, and tell them to use nodejs or other technologies. R and python I haven't run across yet

hyzypg [2:43 PM]
Good to know, thanks for doing this AMA!

daattali [2:44 PM]
no problem its fun!

shoglund [2:46 PM]
Do you ever end up building data processing pipelines for clients using R?

todd [2:46 PM]
@daattali can you think of an example where a client had a need that was better fulfilled by node? i'm curious what kind of problems would lead them to think they need R but in fact need something like node

daattali [2:47 PM]
only if there's something complex about it. Usually my clients are people who do know R and come to me for help with advanced features. I'm really limiting myself in that sense, because I can have more clients with easier jobs..... but that's what I do for now

when somebody wants a fully fledged web service, with multiple pages, connections to databases, hundreds of simultaneous users, etc, and they come to me because some of their analysis uses a few lines of R code, then I often tell them it's not the best tool for the job. It can be done, but would be a waste of their resources

full fledged*

shoglund [2:49 PM]
got it

zackburt [2:51 PM]
:slightly_smiling_face: fyi, I’m sure people in this channel would love to assist with webapp jobs that aren’t necessarily r-oriented, and they’d be happy to do so as subcontractors under your consulting umbrella

daattali [2:52 PM]
that's true, haven't thought of that

will keep in mind for the future

zackburt [3:19 PM]
I would like to thank Dean for this Awesome AMA!

daattali [3:21 PM]
thank you :slightly_smiling_face:

Thanks for visiting this AMA transcript. Please feel free to click around this website and learn more about our community, Code For Cash.

Keep tabs on Code For Cash.

Download the free article on Legal Ideas for freelance software development, and then join our mailing list.
Get our articles on "Finding Clients" and "Communication" for free.