diff --git a/README.md b/README.md
index 2d2f598..3697946 100644
--- a/README.md
+++ b/README.md
@@ -198,32 +198,7 @@ Press **Train** — the network trains in real time, showing:
 * Accuracy
 * Six random preview samples with predicted labels
 
----
 
-## Why This Project Matters
-
-This repository demonstrates:
-
-* understanding of GPU buffer layouts
-* compute pipeline dispatching
-* custom WGSL kernels for ML
-* manual backprop and gradient math
-* GPU memory management and buffer resizing
-* a full deep-learning training loop without any frameworks
-
-It is strong evidence of skill in:
-
-**AI systems engineering**,
-**WebGPU compute design**,
-**neural-network internals**,
-and **performance-oriented GPU development**.
-
-These skills are directly relevant for:
-
-* AI inference runtimes
-* GPU acceleration teams
-* Web-based model execution
-* systems-level optimization roles
 
 ---