Used DVDs sprawled out, held in a hand. These include May (2002), The Fury (1978), and Sideways (2004)

Movie haul from the used book store near my work. Havent seen any of these but they've been on my watchlist for a long while 📽️

"hey can you pull some data for me real quick? I want to report that this number went up in Q1."

Good luck with that buddy, that number went down

I used to joke post-pandemic that things feel mostly the same but everyone got a little weirder in quarantine. This kind of feels that way but now I'm the only one at work that got weirder.

I've basically been at home since November helping my partner heal from being struck by a car. He's doing much better now and is able to do most things independently. It's weird to be getting back in the world but holy shit I needed it.

Taxes done 👍👍

I like Starfield a ton but the inventory system could use a bit of love. Extra filters or something would go a long way.

Experimented tonight with incorporating RAG'd ChatGPT into Slack via Slack App and it works pretty well. My hope is to have this be a front line of contact for users who have process questions about our team's function. Users who want data still need to talk to a human.

I do not know what these cords do so they get put into the box of cords

My first Quarto templated report is going out next week. It's kind of great. Report writing will still have some manual funk but at least the repeatable stuff is taken care of.

Maybe a hot take but I think if you restrict a task to a fairly small data frame, provide it with definitions of the fields, and don't ask it to do math, then AI summaries are not horrible.

Tbh one of the first things I did was give an example data frame to chatgpt and ask it to generate a report of the data. It generated a pdf report with a decent enough summary of the movement of data (number went up significantly!) but it did a very bad job at data visualization.

Unsure how well it'll play in a more automated / airflow workflow, that might be painful. But for semi frequent tasks that will require human written analysis, it's not too bad.

I experimented in the past with some python packages to do this but they all seemed a bit under powered for the task. Quarto seems to work pretty well. Config on the file is light and markdown + python combo is what I needed.

Experimenting with Quarto to automate some of our monthly report building. I'm being asked to generate a monthly report on metrics that matter. Right now it is Excel pivot chart -> word doc heavy. It's too much time fussing with formatting and redoing work I've done in prior months.

Was doing a monthly pull of numbers earlier today and found out the people who owned the source data were all laid off in early Feb.

Outlook's UI is god awful these days

Now to do the stuff I'm very familiar with. Documenting the tables, mapping how they fit together, and making scrappy v1 dashboards.

Full schema rewrite and full data refresh every day. That'll probably be a mistake at some point. At the very least Ill have to change it when incremental loads/upserts are needed.

I don't manually wanna write the create statements for each of these Salesforce objects. i wrote a function to look at the pandas data types, examine the properties of the data, and choose a redshift data type and byte size that's valid.

Data is pulled into panda and cleaned thoroughly. Pandas is funky when converting data types to parquet and redshift hates how they translate to each other, it seems.

Used simple_salesforce's bulk API to pull the data. Our team is still pretty small so full refreshes daily have us well below our api limits. The bulk api doesn't allow select * style queries so I first have to obtain all queryable columns then dynamically generate the select statement.

Salesforce data finally in Redshift! For a first data engineering project I think it went pretty well.

I absolutely love this gay couple on Neighbors ep 2

It's so embarrassing for a journalist to reach the point in their career to be able to ask direct questions to the president and you use that opportunity to stroke his ego.

Just realized what getattr is in python. All this time I've been coding around it.

Can't have the dogs out at the desk!

Metrics become intertwined with company culture. Surveillance changes behavior and the way we relate to one another. Everyone is trying to get the number up and pressure is applied to get it there.

I'm a decent analyst. I've been around data enough to know that it doesn't speak for itself. It is created and shaped into something that allows us to read meaning in it. This translation effort has to be done with some mathematical rigor, historical understanding, and a healthy dose of empathy.

I've been reflecting a lot on how the data team is responsible for shaping the narrative around what is going right and wrong and what the path is forward.

My org is undergoing very public layoffs with "following the data" given as a justification at every turn. As someone working in data at the org, it's laughable.