@doinkpatrol
@CaitCamelia
I always imagine it's because they're having a hard day and could use the extra tip. That was always the case when I was having an off day as a server.
Slack is truly awful for productivity -- It's designed to both disrupt flow-state AND promote impulsive ideas instead of well-considered proposals. I don't know how so many people have been bamboozled into thinking it's a replacement for email.
The takes on the SVB situation are just so incredibly bad. All of society works better if people can trust that their money is safe in a bank.
It’s bananas that that is controversial.
I'm hiring for lots of
#rlang
roles at Recast.
If you're a software engineer specializing in R, we've got a great role for you. Data analyst? Statistician / Data scientist? Please come work with us!
Seems like Bayesian statistics hasn't been picked up very quickly outside of academia. Anyone has thoughts as to why?
Just not enough applications? Other approaches too ingrained in businesses already? People hate uncertainty?
Statistical modeling is hard. I often miss the days of working in machine learning where I just have to hyper-parameter tune my way to lower predictive error. In the world of inferential statistics there's no AUC metric to guide us and we have to use theory to get anywhere.
For my first programming (in Stata!) I was incredibly proud of myself for writing a program 5000 lines long which was the result of copy and pasting the same operation over and over again because I didn't know what a loop was.
A year and a half ago my friend Claire and I started Analytics Engineers Club to create the training course for analytics engineers we wished we had when we were struggling to teach ourselves the fundamentals of analytics engineering.
Probability and statistics are so counter-intuitive for the majority of the population that it makes it really difficult to build any type of product that is both statistically rigorous and usable by non-statisticians.
I spent the beginning of my career building sophisticated algorithms to back into the answer my boss was looking for.
"Analyst degrees of freedom" is one of the most important but least discussed topics in business stats
Once you've seen it, it's hard to trust an analysis again
Thrilled that Locally Optimistic slack for data folks is now over 3k people! But we've got 800 weekly active users and our free plan only lasts about 2 weeks.
We are open to all and have no way to monetize, so if anyone knows someone
@SlackHQ
who can pull strings plz help!
2022 has been a wild year in the venture markets. Many people have said that it's one of the worst times to raise venture money they've ever seen.
And yet.
I'm very stoked to announce that Recast recently closed our seed funding round led by
@LererHippeau
.
@buccocapital
The right thing for society is bad for homeowners. People want to lower the cost of housing without impacting homeowners’ largest financial asset which is obviously impossible so you see people talking like this.
Sometimes I feel like my job is just repeating the question “how can we break this effort up into smaller pieces we can ship faster?” over and over and over again
Incrementality is really the only thing marketers should care about.
Marketers get distracted by platform metrics (impressions, clicks, ROAS) that aren't *really* measuring incrementality. Sometimes they're directionally correct. Other times they're misleading vanity metrics.
📢 Calling all Bayesians 📢
We're opening up another role for a senior data scientist to join our team at Recast.
You'll get to use
@mcmc_stan
to solve some of the toughest problems in statistics powered by our massive cloud-compute infra.
Join us.
We had our first real
@AEng_Club
class today reviewing command line and git. I'm so FIRED UP right now about how great our students are and all of the learning that's happening. It's thrilling!
The most engergy-sucking tasks are the ones where you know there are no good solutions because you get to the end of the task and you're like "well, I did a good job but everything about this still sucks".
Just published a blog post about the power of using Bayesian methods (with example in PyMC3) for doing media mix modeling.
This is a great easy introduction to the power of MCMC if you've been looking for an on-ramp!
@bennstancil
I think for the bulk of analytics work (reporting) SQL is strictly better than python (definitely) and R (mostly). Once there's more modeling or complicated statistical analysis moving to R or python makes sense, but that's a very small % of the work.
‘Airbnb has reported its most profitable fourth quarter ever, two years after slashing its overall marketing investment but shifting spend from performance channels into brand building.’
Enough said.
👋 Recast is growing like crazy and we're looking for a Senior Data Scientist to join our team.
Fluency with R and a passion for causal inference and Bayesian statistics are required.
Strong disagree. A data artifact is "in production" if it's being used to make real decisions or impacts real customers.
That finance excel spreadsheet? Production.
Query on someone's desktop for the board deck reports? Production.
@bennstancil
A data artifact is "in production" if a) it has a SLA and data quality checks that alert when they are violated, and b) all of its upstream sources have a data contract (which I define as a validated schema + data quality checks) which *also* alert when they are violated.
@sethrosen
It's probably the most impactful piece of software ever written. It's taught billions of people how to write software. It's the original low-code / no-code tool.
Every piece of software should strive to be as flexible and easy to use as Excel.
One year ago Recast had zero employees and now we have 8 and are growing fast.
No matter how many blog posts you read about how quickly things change in a high-growth startup it's impossible to appreciate just how wild a ride it is until you actually do it.
I've been doing a lot of interviewing for Recast and it's really driven home the point that "data scientist" is a totally uninformative title.
The breadth of skills and focus areas is so incredibly wide that it's basically uninformative
Hey hey! Applications close TOMORROW for analytics engineers club (
@AEng_Club
) spring cohort.
If you've been looking to go from data analyst -> analytics engineer, this is your chance!
This weekend my gf asked how investors determine the value of a business and about 20 minutes into the spreadsheet lesson on net present value calculations I can’t help but imagine she regretted it
Just booked my first real not-bringing-a-laptop vacation in years and I am incredibly fired up to sit on the beach and just let my eyes unfocus on the horizon for three straight days.
@hspter
When I worked at Analysis Group we did analysis replications. Analyst 1 writes down all of the assumptions, and analyst 2 re-does the analysis from scratch only looking at the assumptions (not the code).
Training up junior folks is very rewarding but getting to work with a team of extremely experienced professionals who are focused on executing at a very high level is just SO MUCH FUN.
As a founder I love working with consultants/contractors before making a full-time hire because that way I can start to see what good (and bad, sometimes) looks like and use that in the interview process. Can really help to avoid painful hiring mistakes.
My two biggest learnings doing startup sales:
1. Every question from a prospect should become a blog post
2. Record every sales call so you can rewatch it later
My gf thinks I'm insane but there's truly nothing I love more than getting up early on a Saturday and running errands and checking things off the todo list
@sethrosen
The trick is to build a product with the right amount of code for the right audience. Looker has lookml for the analysts / engineers, but business users don't need to know anything about it. That was a great design choice.
This discourse drives me crazy. If we don't want investment companies buying up single-family homes, the solution is to make them *bad investments*.
This is very easy to do: just build a lot more housing.
@pdrmnvd
Definitely -- though in general it's a star schema underneath and the OBT is just a "view". As long as end-users have a good way to find the columns they need it prevents bad-join errors and makes their SQL dead simple
Lifehack: order two large pizzas for yourself on Friday night and then eat nothing but pizza for the next three days and then use your lethargy as a great excuse not to exercise.
👋 Hi
#rstats
folks -- we are once again hiring for an R-focused software engineer. We run R in production so are looking for folks with experience maintaining / developing CRAN / Bioconductor packages.
Experience with cloud-first development (AWS) and distributed systems a plus
The Superbowl in the US was last month, and you probably saw loads of advertisements for hot wings leading up to the big game.
Unfortunately, the common advice around how to do marketing analytics would tell you that your marketing is *least effective* around the big game. a 🧵
@ericalouie
splitting the work into two titles allows both to be great. I welcome the return of the analyst as a super highly compensated role building on the infra built by analytics engineers.
It is somewhat bitterweet to say I will be leaving
@zapier
to start a new adventure as a senior product data scientist at
@getrecast
.
Look at me, mom! I'm a professional Bayesian!
📢 Awesome role at Recast alert 📢
We're hiring a junior data scientist to join our outstanding DS team. This is a great role for someone with strong R skills looking to learn more statistics. HMU if you know someone that might be a good fit
A quick explainer on Bayesian Stats that we use a lot in our sales process at Recast:
If you took a statistics class in college, you probably focused on frequentist statistics – things like p-values and t-tests.
Recast just submitted a request to increase our AWS account limit for concurrent EC2 VCPUs to 8,192 if you want a sense of how things are going over here.
Great myth-busting on Facebook's onboarding KPI which many people have pointed out to me as an example of "how we should use data to generate insights".
No data needed! It was obvious that more friends in less time was better so they just picked some numbers!
Woman in the security line loudly arguing with the TSA prson about having her $50 face cream confiscated while simultaneous trying to apply the contents of the entire tube of $50 facecream to her face.
Sometimes I have calls with people who are in a noisy open plan office and I just can’t imagine how anyone gets any deep work done in that environment at all.
Open plan offices are the biggest self-own in the technology world by far.
After a few hundred hours of practice, I'm finally getting to the point where there are moments when my guitar playing actually sounds good.
Learning to just grind through the slog when developing a new skill has been the biggest unlock of my adult life.
Just hired our first salesperson at Recast. Finally seeing what "good" actually looks like.
It's as if I had learned to play the guitar from reading books and then going to a Mark Ribot concert.
Worked on the Recast hiring plan today and under the "ASAP" column we have enough people to double the size of the whole company.
Hold onto your hats folks this is going to be a wild ride.
👋 hi! I'm hiring at Recast for our first full-time sales hire. We've seen incredible growth over the last year and we're looking to add an awesome account executive to the team. Looking for
1) Experience with ACV from 75k - 150k
2) Startup experience
3) Martech experience
@PhDemetri
These sorts of questions generally are about knowing a bunch of formal probability "tricks" which are useful in some very thin domain slices but in general with any problem like this I'd want an employee to *start* with a simulation or a dgp not start with formulas
I keep replaying over and over in my head how badly I screwed up yesterday when I came into the dining room and caught the dog fully on top of the table with a mouthful of turkey and I yelled at him to get down instead of recording a video and going viral :(
At
@getrecast
people often ask us about what "Bayesian" means and why it's so important for doing marketing mix modeling correctly.
Talk of bayes theorem and priors and posteriors might be helpful for statitisticians, but today I'll try to give a more intuitive explanation. 1/7
This week,
@hammer_mt
and I finally published our epic piece on marketing measurement in
@lennysan
's newsletter.
This project took months of effort to collect all of the case studies and then craft a narrative around how to bring all the marketing measurement pieces together
@gwenwindflower
@sarahcat21
I honestly don't get this. It seems obvious to me that self-service delivers tons of value. When did everyone decide against that?
Huge congrats to Rok Cesnovar on his successful PhD thesis defense (all about parallel computing in
@mcmc_stan
)!
It's such a pleasure to get to work with you every day at
@getrecast
, Rok!
@drewbanin
@aerialfly
@josh_wills
IDK I have worked as a data / analytics engineer and as a product owner and I can tell you right now that I weight faster product movement over good data / analytics by like 10 to 1. In many contexts slowing down product velocity for data reasons is a bad tradeoff
Putting process in place in a startup is a real tightrope walk:
too much and you strangle your dynamism. too little, and everything blows up from disorganization.
About two months ago, I adopted a dog. His names is Rulfo (for Juan Rulfo). He's wonderful, but adopting Rulfo was WAY more work than I expected, so I figured I'd share a bit about what I learned.
Thanks to this we've been able to lower the tuition costs and get the content in the hands of way more students.
If you're a data analyst looking to make the transition into analytics engineering, I highly recommend you check it out!
Sometimes it can seem like there's an immense chasm between "junior" and "senior" performers in a startup.
There are a few patterns of behavior that "seniors" do consistently that I figured I'd share in case it's helpful for those looking to grow in their careers.
Question: what are the most important non-language-specific practical concepts for a junior software engineer to be familiar with?
Testing? Debugging? Version control?
Thinking through this for
@AEng_Club
and also just in general.
Have been a huge fan of
@eric_seufert
for a long time so I was really, really excited to get to chat with him about probabilistic measurement for the MDM podcast.
This week's
@MobileDevMemo
podcast episode features an interview with
@Mike_Kaminsky
, the CEO of Recast. We discuss probabilistic marketing attribution, including: (1/X)
Day 4 of my NYC trip and while I'm once again fired up by the energy of the city I am exhausted by the constant stream of dinners and activities.
How do you people live like this? I'm feeling in desperate need of a good four-hour wall stare.
At Recast, we've often found that just getting results from an MMM is not that hard. What *is* hard is actually taking the MMM and using it to drive business value.
I've spent the last few months working on writing a 50+ page ebook on how modern brands can actually *use* MMM.
@kierisi
In Statistical Rethinking
@rlmcelreath
often says things like "if you find that explanation confusing, it just means you're paying attention" which I found quite comforting.
For all my MMM nerds, we just published our write up of the google-affiliated LightweightMMM open-source project:
Overall seems like a very great starting place for those wanting to dip their toes in Bayesian MMM