diff --git a/content/courses/parallel-computing-introduction/distributed_mpi_memory_layouts.md b/content/courses/parallel-computing-introduction/distributed_mpi_memory_layouts.md
index f83bca0b..fbcdc178 100644
--- a/content/courses/parallel-computing-introduction/distributed_mpi_memory_layouts.md
+++ b/content/courses/parallel-computing-introduction/distributed_mpi_memory_layouts.md
@@ -15,7 +15,7 @@ As we learned, MPI does not construct buffers from variables; it reads a specifi
 
 In a given programming language, an array can be _row major oriented_ or _column major oriented_.  In a row-major oriented language (most of them, including C/C++ and Python) a two-dimensional array is represented in memory in terms of its indices as
 
-{{< diagram >}}
+{{< diagram alt="Diagram showing a 4×4 two-dimensional array stored in row-major order, where elements are laid out row by row and the second index changes fastest in memory." >}}
 flowchart LR
    A[0,0] --- B[0,1]
    B --- C[0,2]
@@ -40,7 +40,7 @@ In C/C++ the rows may or may not be contiguous in memory, but the elements of ea
 
 In a column-major oriented language (Fortran and some others such as Matlab, Julia, and R) the layout is by column. These languages also count from 1, at least by default, as represented below.
 
-{{< diagram >}}
+{{< diagram alt="Diagram showing a 4×4 two-dimensional array stored in column-major order, where elements are laid out column by column and the first index changes fastest in memory." >}}
 flowchart LR
    A[1,1] --- B[2,1]
    B --- C[3,1]
diff --git a/content/courses/parallel-computing-introduction/parallel_basics.md b/content/courses/parallel-computing-introduction/parallel_basics.md
index 46536781..855f5b22 100644
--- a/content/courses/parallel-computing-introduction/parallel_basics.md
+++ b/content/courses/parallel-computing-introduction/parallel_basics.md
@@ -79,7 +79,7 @@ The landscape service has several tasks to perform.   
 
 We can draw a graph to help us understand the sequence of events.
 
-{{< diagram >}}
+{{< diagram alt="Diagram showing landscaping workflow: turn off security system, perform mowing, edging, and weeding in parallel, set sprinklers, then turn security system back on." >}}
 graph TD;
 A(Turn off Security System) --> B(Edge Lawn)
 A(Turn off Security System) --> C(Mow Lawn)
diff --git a/content/courses/python-high-performance/serial-optimization.md b/content/courses/python-high-performance/serial-optimization.md
index fa513084..427afd5f 100644
--- a/content/courses/python-high-performance/serial-optimization.md
+++ b/content/courses/python-high-performance/serial-optimization.md
@@ -11,7 +11,7 @@ menu:
 
 We can represent the optimization process with a flowchart:
 
-{{< diagram >}}
+{{< diagram alt="Flowchart of the optimization process: profile code, tune the slowest section, check for performance improvement, then repeat or try a different solution." >}}
 graph TD;
 A(Profile or time) --> B(Tune the slowest section);
 B --> C{Performance increase?};
@@ -20,7 +20,7 @@ C -- No --> E(Try a different solution);
 D --> A
 {{< /diagram >}}
 
-There are many approaches to speeding up sections, some specific to Python and some more generic.  In this section we will consider several possibilities.
+There are many many approaches to speeding up sections, some specific to Python and some more generic.  In this section we will consider several possibilities.
 
 ### Avoid for Loops