diff --git a/dev/cuda/Makefile b/dev/cuda/Makefile
index a0ba17a..844c46c 100644
--- a/dev/cuda/Makefile
+++ b/dev/cuda/Makefile
@@ -1,88 +1,63 @@
-NVCC := $(shell which nvcc 2>/dev/null)
+# Makefile for building dev/cuda kernels
 
+# Find nvcc
+NVCC := $(shell which nvcc 2>/dev/null)
 ifeq ($(NVCC),)
 		$(error nvcc not found.)
 endif
 
-TARGETS = adamw attention_backward attention_forward classifier_fused crossentropy_forward crossentropy_softmax_backward encoder_backward encoder_forward gelu_forward layernorm_backward layernorm_forward matmul_backward matmul_backward_bias matmul_forward nccl_all_reduce residual_forward softmax_forward trimat_forward
-
+# Compiler flags
 CFLAGS = -O3 --use_fast_math
+MPI_PATHS = -I/usr/lib/x86_64-linux-gnu/openmpi/include -L/usr/lib/x86_64-linux-gnu/openmpi/lib/ 
+
+
+%: %.cu
+	$(NVCC) $(CFLAGS) $< -o $@ -lcublas
+
+TARGETS = adamw attention_backward attention_forward classifier_fused crossentropy_forward crossentropy_softmax_backward encoder_backward encoder_forward gelu_forward layernorm_backward layernorm_forward matmul_backward matmul_backward_bias matmul_forward nccl_all_reduce residual_forward softmax_forward trimat_forward
 
 all: $(TARGETS)
 
-adamw: adamw.cu
-	$(NVCC) $(CFLAGS) adamw.cu -o adamw
-
-attention_backward: attention_backward.cu
-	$(NVCC) $(CFLAGS) attention_backward.cu -o attention_backward -lcublas
+# Forward kernels
 
 attention_forward: attention_forward.cu
-	$(NVCC) $(CFLAGS) attention_forward.cu -o attention_forward -lcublas
-
 classifier_fused: classifier_fused.cu
-	$(NVCC) $(CFLAGS) classifier_fused.cu -o classifier_fused
-
 crossentropy_forward: crossentropy_forward.cu
-	$(NVCC) $(CFLAGS) crossentropy_forward.cu -o crossentropy_forward
-
-crossentropy_softmax_backward: crossentropy_softmax_backward.cu
-	$(NVCC) $(CFLAGS) crossentropy_softmax_backward.cu -o crossentropy_softmax_backward
-
-encoder_backward: encoder_backward.cu
-	$(NVCC) $(CFLAGS) encoder_backward.cu -o encoder_backward
-
 encoder_forward: encoder_forward.cu
-	$(NVCC) $(CFLAGS) encoder_forward.cu -o encoder_forward
-
 gelu_forward: gelu_forward.cu
-	$(NVCC) $(CFLAGS) gelu_forward.cu -o gelu_forward
-
-layernorm_backward: layernorm_backward.cu
-	$(NVCC) $(CFLAGS) layernorm_backward.cu -o layernorm_backward
-
 layernorm_forward: layernorm_forward.cu
-	$(NVCC) $(CFLAGS) layernorm_forward.cu -o layernorm_forward
+matmul_forward: matmul_forward.cu
+	$(NVCC) $(CFLAGS) -Xcompiler -fopenmp matmul_forward.cu -o matmul_forward -lcublas -lcublasLt
+residual_forward: residual_forward.cu
+softmax_forward: softmax_forward.cu
+trimat_forward: trimat_forward.cu
 
+# Backward kernels
+
+attention_backward: attention_backward.cu
+crossentropy_softmax_backward: crossentropy_softmax_backward.cu
+encoder_backward: encoder_backward.cu
+layernorm_backward: layernorm_backward.cu
+matmul_backward_bias: matmul_backward_bias.cu
 matmul_backward: matmul_backward.cu
 	$(NVCC) $(CFLAGS) -Xcompiler -fopenmp matmul_backward.cu -o matmul_backward -lcublas
 
-matmul_backward_bias: matmul_backward_bias.cu
-	$(NVCC) $(CFLAGS) matmul_backward_bias.cu -o matmul_backward_bias
+# Update kernels
 
-matmul_forward: matmul_forward.cu
-	$(NVCC) $(CFLAGS) -Xcompiler -fopenmp matmul_forward.cu -o matmul_forward -lcublas -lcublasLt
+adamw: adamw.cu
+
+# NCCL
 
 nccl_all_reduce: nccl_all_reduce.cu
-	$(NVCC) -lmpi -lnccl -I/usr/lib/x86_64-linux-gnu/openmpi/include -L/usr/lib/x86_64-linux-gnu/openmpi/lib/ nccl_all_reduce.cu -o nccl_all_reduce
+	$(NVCC) -lmpi -lnccl $(MPI_PATHS) nccl_all_reduce.cu -o nccl_all_reduce
 
-residual_forward: residual_forward.cu
-	$(NVCC) $(CFLAGS) residual_forward.cu -o residual_forward
-
-softmax_forward: softmax_forward.cu
-	$(NVCC) $(CFLAGS) softmax_forward.cu -o softmax_forward
-
-trimat_forward: trimat_forward.cu
-	$(NVCC) $(CFLAGS) trimat_forward.cu -o trimat_forward -lcublas
+run_all: all
+	@for target in $(TARGETS); do \
+		echo "\n========================================"; \
+		echo "Running $$target ..."; \
+		echo "========================================\n"; \
+		./$$target; \
+	done
 
 clean:
 	rm -f $(TARGETS)
-
-run_all: all
-	./adamw
-	./attention_backward
-	./attention_forward
-	./classifier_fused
-	./crossentropy_forward
-	./crossentropy_softmax_backward
-	./encoder_backward
-	./encoder_forward
-	./gelu_forward
-	./layernorm_backward
-	./layernorm_forward
-	./matmul_backward
-	./matmul_backward_bias
-	./matmul_forward
-	./nccl_all_reduce
-	./residual_forward
-	./softmax_forward
-	./trimat_forward
diff --git a/dev/cuda/README.md b/dev/cuda/README.md
index 5668360..89d8504 100644
--- a/dev/cuda/README.md
+++ b/dev/cuda/README.md
@@ -1,3 +1,7 @@
 # dev/cuda
 
-This directory is scratch space for developing various versions of the needed CUDA kernels. Each file develops a kernel, see the top of each file for instructions on how to compile and run each one.
+This directory is scratch space for developing various versions of the needed CUDA kernels. Each file develops a kernel, see the top of each file for instructions on how to compile and run each one using the `nvcc` compiler.
+
+An alternative to invoking `nvcc` manually is to use `make` with the accompanying `Makefile` in this directory. Each kernel has its own `make` build target, invoking `make` for the target builds the associated binary. 
+
+For example, `make gelu_forward` builds the forward GELU kernel, creating a binary that can be executed by running `./gelu_forward`. `make` or `make all` builds all the kernels in this directory. To delete all binary build targets, run `make clean`.