AI integration:

 - refactor using Kiro/Claude to bring in AI agent ideas

Author: mnfienen

AGENDA.md

@@ -4,9 +4,9 @@
 Monday
 - 900-930: Introductions and getting started  (Mike) 
 - 930-1000: Python overview and Quick orientation to jupyter, environments, and mamba/conda (Andy) 
-- 1000-1130: 00 Python basics review and AI overlords: Q&A and discussion https://cscircles.cemc.uwaterloo.ca/  (A: word counting ) (Mike) 
+- 1000-1130: 00 Python basics review and AI overlords: Q&A and discussion https://cscircles.cemc.uwaterloo.ca/  (A: word counting ) (Mike) — *includes AI Foundations Notebook (00b_ai_assisted_programming.ipynb): overview of AI coding assistants, strengths/limitations, worked failure examples, prompt patterns, and responsible AI use*
 - 1130-1230: lunch
-- 1230-1330: 01 Functions and Scripts  (A: build a function) (Kalle) 
+- 1230-1330: 01 Functions and Scripts  (A: build a function) (Kalle) — *includes AI Sidebar*
 - 1330-1345: break
 - 1345-1545: 02 Namespace, Modules, Packages, Objects (A: import and use own functions, etc) (Josh) 
 - 1545-1600: break
@@ -16,33 +16,33 @@ Monday
 
 ## Tuesday
 - 900-930:  04 Working with files, string formatting (Nick) (continued)
-- 930-1130: 05 Numpy (where, slicing, indexing)(A: Mt. St. Helens) (Mike)
+- 930-1130: 05 Numpy (where, slicing, indexing)(A: Mt. St. Helens) (Mike) — *includes AI Sidebar*
 - 1130-1230: lunch
-- 1230-1430: 06 Matplotlib (A: make a semi-complex plot, animation, PdfPages, USGS Style) (Kalle)
+- 1230-1430: 06 Matplotlib (A: make a semi-complex plot, animation, PdfPages, USGS Style) (Kalle) — *includes AI Sidebar*
 - 1430-1445: break
-- 1445-1615: 07 Theis exercise + VSCode IDE debugging, liveshare (Andy) 
+- 1445-1615: 07 Theis exercise + VSCode IDE debugging, liveshare (Andy) — *includes AI Sidebar*
 - 1615-1630: break
-- 1630-1730: 08 Pandas (interesting data, viz, excel, apply correction to data, indexing, grouping, retrieving data from NWIS (dataretrieval) (Josh) 
+- 1630-1730: 08 Pandas (interesting data, viz, excel, apply correction to data, indexing, grouping, retrieving data from NWIS (dataretrieval) (Josh) — *includes AI Sidebar*
 - 1730: Class Adjourn
 
 ## Wednesday
 - 900-1000: 08 Pandas (interesting data, viz, excel, apply correction to data, indexing, grouping, retrieving data from NWIS (dataretrieval) (Josh) (continued)
-- 1000-1130: 09 GeoPandas for open-source GIS and viz (Mike)
+- 1000-1130: 09 GeoPandas for open-source GIS and viz (Mike) — *includes AI Sidebar*
 - 1130-1230: lunch
 - 1230-1300: 09 GeoPandas for open-source GIS and viz (Mike) (A: working with open-source city data) (continued)
 - 1300-1430: 10 Rasterio (Nick) 
 - 1430-1445: break
 - 1445-1615: 11 Xarray/rioxarray (Andy) 
 - 1615-1630: break
-- 1630-1730: Code style/quality, testing, version control (Andy and all)
+- 1630-1730: Code style/quality, testing, version control (Andy and all) — *includes Verification Notebook (12_verifying_ai_code.ipynb): verifying AI-generated code using analytical solutions, unit testing, and dimensional analysis*
 - 1730: class adjourn
 
 # Part 1: Application of Python and Flopy to Groundwater Modeling 
 ## Thursday
-- 900-1000: 01 Introductions and IT sorting (Mike)
-- 1000-1100: 02 Build your first model with FloPy (Nick)
+- 900-1000: 01 Introductions and IT sorting (Mike) — *includes AI Sidebar*
+- 1000-1100: 02 Build your first model with FloPy (Nick) — *includes AI Sidebar*
 - 1100-1115: break
-- 1115-1245: 03 Loading and visualizing models with FloPy (Andy) 
+- 1115-1245: 03 Loading and visualizing models with FloPy (Andy) — *includes AI Sidebar*
 - 1245-1345: lunch
 - 1345-1615: 04 Intersection Capabilities, modelgrid, time wrangling (Josh) 
 - 1615-1630: break
@@ -60,3 +60,12 @@ Monday
 - 1600-1630: 09 Run and process MF6T (Kalle) 
 - 1630-1700: 10 Run and process MODPATH/MF6-PRT models (Andy) 
 - 1700: Wrap up
+
+---
+
+## Optional Self-Study Materials
+
+The following AI-integrated notebooks are available for self-study during breaks or after hours:
+
+- **Prompt Engineering Notebook** (00c_prompt_engineering.ipynb): Five prompt patterns for scientific computing, comparative exercises, iterative refinement demos, and physical constraints exercises
+- **AI Capstone Exercise** (13_ai_capstone.ipynb): Multi-step hydrology data analysis combining manual coding and AI assistance (download NWIS data → process with Pandas → create publication-quality plot)

admin/mike_ai_notes.md

@@ -0,0 +1,18 @@
+# 00b
+- the baseflow example is a little precious - there must be something better
+
+# 00c
+- The "verification" for Pattern A is bad - by concluding the calculation of the mean is 5.0 from a series of all the value 5.0 misses potential errors like basically any other function of the data (min, unique, max, median, etc.)
+- Pattern B is a little better
+- Pattern C - makes a good point about specificity, but maybe assumes _too much_ prior knowledge of the functionality
+- Pattern D - so, this works, but pretty gross and compact
+- The Part 2 Vague prompt/structured prompt example makes a good point but stupid way to provide the data. Can do better than that!
+- Physical constraints answer - still doesn't use T in the final answer - junkus! Doesn't actually need it, but doesn't say so. Junkus.
+
+# 06 matplotlib
+- first example is silly with synthetic data - better to rewrite using real data but then offloading the formatting syntax stuff to AI
+- we should refactor this entire sidebar to consider just the formatting question. The AI fails here are trivial and not very interesting
+
+# 12 verification
+- first example is kinda comical because it's suggesting we should verify AI with an analytical solution. Which is a generally good idea, but if it's an analytical solution to start with, this is circular
+-