Who is an analytics engineer? - Transform data in your warehouse

class: title, center, middle
# Who is an analytics engineer?

---
# The traditional data team

.left-column-33[
#### Data Engineers
- Build data warehouse infrastructure
- Extract & load data into your warehouse
- Transform data to make it easier to work with
]
.right-column-66[
.left-column[
#### Data Analyst
- Build dashboards & reports
- Insights work
]
.right-column[
#### Stakeholder
- Politely wait for their analyst to give them numbers
]
]

???
Slide setup: When did people first start hearing the term 'analytics engineering'? Analytics Engineering is relatively new. The role of Analytics Engineer is relatively new. To understand it, we need to understand how data teams traditionally looked.

Note: Lots of custom engineering work. Distance between business stakeholders and those doing the actual transformation of their data.

---
# Modern data stack

???
- Reliable out of the box data loader tools like Stitch and Fivetran made it easy to get data into your warehouse, but the data is in a source conformed format.
- MPP (massively parallel processing) warehouses such as Snowflake, Redshift, and BigQuery replaced the need for engineers to maintain warehouse infrastructures.
- BI tools (Looker, Mode, and others) allowed analysts and consumers to do their own deep dives into data.
- The combination of these tools is what we call the modern data stack and it has created huge improvements for data teams however with great power comes great responsibility. 
- Problem: people only had access to the raw data

Teaching notes:
- flip back to traditional data team and annotate to show how it doesn't make sense anymore (cross parts out, arrow other parts around, particularly stakeholders not wanting to wait for the analysts)
---
# Modern data stack: consequences
.dense-text[
Have you ever:
- Had two stakeholders have different values for a KPI, and not been sure what was right?
- Rewritten the same snippet of SQL? Or forgotten to write it?
```sql
select * from orders
where status not in ('cancelled', 'false')
    and deleted_at is null
    and user_id != 12
```
- Broken all your dashboards because of a bad join?
- Had data go stale? And only found out when someone else noticed?
- Not really been sure where your data came from?
]
???
audience: Share your story of the silliest data problem you've had
---
# Enter: dbt

.center[_Read the [Viewpoint](https://docs.getdbt.com/docs/about/viewpoint/)_]

???
- At it’s core, dbt facilitates the application of software engineering best practices to analytics work - things like version control, logging, alerting, automated testing and deployment. 
- Enter: dbt. At it’s core, dbt facilitates the application of software engineering best practices to analytics work - things like version control, logging, alerting, automated testing and deployment. 
- Anyone who knows SQL can do this!
- This role is neither data engineering nor data analysis... it's analytics engineering

---

## dbt & Analytics Engineering

???
What do these software engineering best practices look like?

All that and more... after the break!

---
class: subtitle
# Questions so far?

---

class: subtitle

# Zoom Out

<li> Distributed dbt Learn norms</li>

<li> Who is an analytics engineer?</li>

--
<li class="orange"> dbt Fundamentals</li>

<li class="grey">Working Group 1</li>

<li class="grey">dbt Project Design</li>

<li class="grey">Working Group 2</li>

<li class="grey">Testing</li>

<li class="grey">Sources</li>

<li class="grey">Deployment</li>

<li class="grey">Working Group 3</li>

---