DEV Community

AI Engineering Field Reports Series' Articles

Back to Michael Truong's Series
Schema first, prompt second: valid JSON wasn't enough
Cover image for Schema first, prompt second: valid JSON wasn't enough

Schema first, prompt second: valid JSON wasn't enough

5
Comments 7
3 min read
One good example beat every AI writing rule I wrote
Cover image for One good example beat every AI writing rule I wrote

Examples carry taste in a compressed form

One good example beat every AI writing rule I wrote

3
Comments 8
3 min read
The agent plan had every step except where to stop
Cover image for The agent plan had every step except where to stop

Branch protection as the only hard stop

The agent plan had every step except where to stop

13
Comments 40
6 min read
The AI reviewer scored 23/25 and missed the point
Cover image for The AI reviewer scored 23/25 and missed the point

Prioritizing analysis over numerical scores

The AI reviewer scored 23/25 and missed the point

8
Comments 21
5 min read
Upgrades don't have to be a blind trust exercise
Cover image for Upgrades don't have to be a blind trust exercise

Upgrades don't have to be a blind trust exercise

7
Comments 4
5 min read
I fixed my AI reviewer. Then I kept solving the wrong problem
Cover image for I fixed my AI reviewer. Then I kept solving the wrong problem

Naming cognitive jobs over rubric tuning

I fixed my AI reviewer. Then I kept solving the wrong problem

12
Comments 30
4 min read
Model experiments became an architectural stress test
Cover image for Model experiments became an architectural stress test

Models masking hidden system flaws

Model experiments became an architectural stress test

9
Comments 4
5 min read
Active players looked real until we asked which sessions counted
Cover image for Active players looked real until we asked which sessions counted

Active players looked real until we asked which sessions counted

16
Comments 15
4 min read