diff --git a/extmod/ulab b/extmod/ulab
index b64fa6d4c7..2aae646485 160000
--- a/extmod/ulab
+++ b/extmod/ulab
@@ -1 +1 @@
-Subproject commit b64fa6d4c73287edef9ccf09cfd6ec5009f9628b
+Subproject commit 2aae6464856e5213e45c72681dbb20c708fb3c6c
diff --git a/locale/circuitpython.pot b/locale/circuitpython.pot
index 146c65f3e1..fb4d6991b1 100644
--- a/locale/circuitpython.pot
+++ b/locale/circuitpython.pot
@@ -929,11 +929,11 @@ msgstr ""
 msgid "Extended advertisements with scan response not supported."
 msgstr ""
 
-#: extmod/ulab/code/fft/fft.c
+#: extmod/ulab/code/numpy/fft/fft_tools.c
 msgid "FFT is defined for ndarrays only"
 msgstr ""
 
-#: extmod/ulab/code/fft/fft.c
+#: extmod/ulab/code/numpy/fft/fft_tools.c
 msgid "FFT is implemented for linear arrays only"
 msgstr ""
 
@@ -2387,11 +2387,11 @@ msgstr ""
 msgid "arg is an empty sequence"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "argsort argument must be an ndarray"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "argsort is not implemented for flattened arrays"
 msgstr ""
 
@@ -2399,7 +2399,7 @@ msgstr ""
 msgid "argument has wrong type"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
 msgid "argument must be ndarray"
 msgstr ""
 
@@ -2412,7 +2412,8 @@ msgstr ""
 msgid "argument should be a '%q' not a '%q'"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "arguments must be ndarrays"
 msgstr ""
 
@@ -2425,11 +2426,11 @@ msgstr ""
 msgid "array/bytes required on right side"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "attempt to get (arg)min/(arg)max of empty sequence"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "attempt to get argmin/argmax of an empty sequence"
 msgstr ""
 
@@ -2437,15 +2438,15 @@ msgstr ""
 msgid "attributes not supported yet"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "axis is out of bounds"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "axis must be None, or an integer"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "axis too long"
 msgstr ""
 
@@ -2746,19 +2747,19 @@ msgstr ""
 msgid "conversion to object"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/numpy/filter/filter.c
 msgid "convolve arguments must be linear arrays"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/numpy/filter/filter.c
 msgid "convolve arguments must be ndarrays"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/numpy/filter/filter.c
 msgid "convolve arguments must not be empty"
 msgstr ""
 
-#: extmod/ulab/code/poly/poly.c
+#: extmod/ulab/code/numpy/poly/poly.c
 msgid "could not invert Vandermonde matrix"
 msgstr ""
 
@@ -2766,15 +2767,15 @@ msgstr ""
 msgid "couldn't determine SD card version"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "cross is defined for 1D arrays of length 3"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/scipy/optimize/optimize.c
 msgid "data must be iterable"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/scipy/optimize/optimize.c
 msgid "data must be of equal length"
 msgstr ""
 
@@ -2811,18 +2812,14 @@ msgstr ""
 msgid "dict update sequence has wrong length"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "diff argument must be an ndarray"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "differentiation order out of range"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c
-msgid "dimensions do not match"
-msgstr ""
-
 #: py/modmath.c py/objfloat.c py/objint_longlong.c py/objint_mpz.c py/runtime.c
 #: shared-bindings/math/__init__.c
 msgid "division by zero"
@@ -2938,11 +2935,11 @@ msgstr ""
 msgid "filesystem must provide mount method"
 msgstr ""
 
-#: extmod/ulab/code/vector/vectorise.c
+#: extmod/ulab/code/numpy/vector/vector.c
 msgid "first argument must be a callable"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/scipy/optimize/optimize.c
 msgid "first argument must be a function"
 msgstr ""
 
@@ -2950,11 +2947,7 @@ msgstr ""
 msgid "first argument must be a tuple of ndarrays"
 msgstr ""
 
-#: extmod/ulab/code/ndarray.c
-msgid "first argument must be an iterable"
-msgstr ""
-
-#: extmod/ulab/code/vector/vectorise.c
+#: extmod/ulab/code/numpy/vector/vector.c
 msgid "first argument must be an ndarray"
 msgstr ""
 
@@ -2966,7 +2959,7 @@ msgstr ""
 msgid "flattening order must be either 'C', or 'F'"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "flip argument must be an ndarray"
 msgstr ""
 
@@ -2999,7 +2992,7 @@ msgstr ""
 msgid "function got multiple values for argument '%q'"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/scipy/optimize/optimize.c
 msgid "function has the same sign at the ends of interval"
 msgstr ""
 
@@ -3074,7 +3067,7 @@ msgstr ""
 msgid "index is out of bounds"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 #: ports/esp32s2/common-hal/pulseio/PulseIn.c py/obj.c
 #: shared-bindings/bitmaptools/__init__.c
 msgid "index out of range"
@@ -3088,7 +3081,7 @@ msgstr ""
 msgid "indices must be integers, slices, or Boolean lists"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/scipy/optimize/optimize.c
 msgid "initial values must be iterable"
 msgstr ""
 
@@ -3108,7 +3101,7 @@ msgstr ""
 msgid "input argument must be an integer, a tuple, or a list"
 msgstr ""
 
-#: extmod/ulab/code/fft/fft.c
+#: extmod/ulab/code/numpy/fft/fft_tools.c
 msgid "input array length must be power of 2"
 msgstr ""
 
@@ -3116,15 +3109,15 @@ msgstr ""
 msgid "input arrays are not compatible"
 msgstr ""
 
-#: extmod/ulab/code/poly/poly.c
+#: extmod/ulab/code/numpy/poly/poly.c
 msgid "input data must be an iterable"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
 msgid "input matrix is asymmetric"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
 msgid "input matrix is singular"
 msgstr ""
 
@@ -3140,23 +3133,23 @@ msgstr ""
 msgid "input must be an ndarray"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/scipy/signal/signal.c
 msgid "input must be one-dimensional"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
 msgid "input must be square matrix"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "input must be tuple, list, range, or ndarray"
 msgstr ""
 
-#: extmod/ulab/code/poly/poly.c
+#: extmod/ulab/code/numpy/poly/poly.c
 msgid "input vectors must be of equal length"
 msgstr ""
 
-#: extmod/ulab/code/poly/poly.c
+#: extmod/ulab/code/numpy/poly/poly.c
 msgid "inputs are not iterable"
 msgstr ""
 
@@ -3168,7 +3161,7 @@ msgstr ""
 msgid "integer required"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/numpy/approx/approx.c
 msgid "interp is defined for 1D arrays of equal length"
 msgstr ""
 
@@ -3257,11 +3250,7 @@ msgstr ""
 msgid "issubclass() arg 2 must be a class or a tuple of classes"
 msgstr ""
 
-#: extmod/ulab/code/ndarray.c
-msgid "iterables are not of the same length"
-msgstr ""
-
-#: extmod/ulab/code/linalg/linalg.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
 msgid "iterations did not converge"
 msgstr ""
 
@@ -3329,7 +3318,11 @@ msgstr ""
 msgid "math domain error"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
+msgid "matrix dimensions do not match"
+msgstr ""
+
+#: extmod/ulab/code/numpy/linalg/linalg.c
 msgid "matrix is not positive definite"
 msgstr ""
 
@@ -3351,15 +3344,15 @@ msgstr ""
 msgid "maximum recursion depth exceeded"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/scipy/optimize/optimize.c
 msgid "maxiter must be > 0"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/scipy/optimize/optimize.c
 msgid "maxiter should be > 0"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "median argument must be an ndarray"
 msgstr ""
 
@@ -3380,7 +3373,7 @@ msgstr ""
 msgid "module not found"
 msgstr ""
 
-#: extmod/ulab/code/poly/poly.c
+#: extmod/ulab/code/numpy/poly/poly.c
 msgid "more degrees of freedom than data points"
 msgstr ""
 
@@ -3498,7 +3491,7 @@ msgstr ""
 msgid "non-zero timeout must be >= interval"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
 msgid "norm is defined for 1D and 2D arrays"
 msgstr ""
 
@@ -3599,8 +3592,8 @@ msgstr ""
 msgid "only slices with step=1 (aka None) are supported"
 msgstr ""
 
-#: extmod/ulab/code/compare/compare.c extmod/ulab/code/ndarray.c
-#: extmod/ulab/code/vector/vectorise.c
+#: extmod/ulab/code/ndarray.c extmod/ulab/code/numpy/compare/compare.c
+#: extmod/ulab/code/numpy/vector/vector.c
 msgid "operands could not be broadcast together"
 msgstr ""
 
@@ -3608,7 +3601,7 @@ msgstr ""
 msgid "operation is implemented for 1D Boolean arrays only"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "operation is not implemented on ndarrays"
 msgstr ""
 
@@ -3753,7 +3746,7 @@ msgstr ""
 msgid "raw f-strings are not implemented"
 msgstr ""
 
-#: extmod/ulab/code/fft/fft.c
+#: extmod/ulab/code/numpy/fft/fft_tools.c
 msgid "real and imaginary parts must be of equal length"
 msgstr ""
 
@@ -3788,7 +3781,7 @@ msgstr ""
 msgid "rgb_pins[%d] is not on the same port as clock"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "roll argument must be an ndarray"
 msgstr ""
 
@@ -3835,7 +3828,7 @@ msgstr ""
 msgid "single '}' encountered in format string"
 msgstr ""
 
-#: extmod/ulab/code/linalg/linalg.c
+#: extmod/ulab/code/numpy/linalg/linalg.c
 msgid "size is defined for ndarrays only"
 msgstr ""
 
@@ -3843,10 +3836,6 @@ msgstr ""
 msgid "sleep length must be non-negative"
 msgstr ""
 
-#: extmod/ulab/code/ndarray.c
-msgid "slice step can't be zero"
-msgstr ""
-
 #: py/objslice.c py/sequence.c
 msgid "slice step cannot be zero"
 msgstr ""
@@ -3859,19 +3848,19 @@ msgstr ""
 msgid "soft reboot\n"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "sort argument must be an ndarray"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/scipy/signal/signal.c
 msgid "sos array must be of shape (n_section, 6)"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/scipy/signal/signal.c
 msgid "sos[:, 3] should be all ones"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/scipy/signal/signal.c
 msgid "sosfilt requires iterable arguments"
 msgstr ""
 
@@ -3985,7 +3974,7 @@ msgstr ""
 msgid "too many arguments provided with the given format"
 msgstr ""
 
-#: extmod/ulab/code/ulab_create.c
+#: extmod/ulab/code/ndarray.c extmod/ulab/code/ulab_create.c
 msgid "too many dimensions"
 msgstr ""
 
@@ -3998,11 +3987,11 @@ msgstr ""
 msgid "too many values to unpack (expected %d)"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/numpy/approx/approx.c
 msgid "trapz is defined for 1D arrays"
 msgstr ""
 
-#: extmod/ulab/code/approx/approx.c
+#: extmod/ulab/code/numpy/approx/approx.c
 msgid "trapz is defined for 1D arrays of equal length"
 msgstr ""
 
@@ -4144,6 +4133,10 @@ msgstr ""
 msgid "value_count must be > 0"
 msgstr ""
 
+#: extmod/ulab/code/numpy/linalg/linalg.c
+msgid "vectors must have same lengths"
+msgstr ""
+
 #: ports/esp32s2/common-hal/alarm/pin/__init__.c
 msgid "wakeup conflict"
 msgstr ""
@@ -4173,7 +4166,7 @@ msgstr ""
 msgid "window must be <= interval"
 msgstr ""
 
-#: extmod/ulab/code/numerical/numerical.c
+#: extmod/ulab/code/numpy/numerical/numerical.c
 msgid "wrong axis index"
 msgstr ""
 
@@ -4181,7 +4174,7 @@ msgstr ""
 msgid "wrong axis specified"
 msgstr ""
 
-#: extmod/ulab/code/vector/vectorise.c
+#: extmod/ulab/code/numpy/vector/vector.c
 msgid "wrong input type"
 msgstr ""
 
@@ -4197,7 +4190,7 @@ msgstr ""
 msgid "wrong operand type"
 msgstr ""
 
-#: extmod/ulab/code/vector/vectorise.c
+#: extmod/ulab/code/numpy/vector/vector.c
 msgid "wrong output type"
 msgstr ""
 
@@ -4221,14 +4214,14 @@ msgstr ""
 msgid "zero step"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/scipy/signal/signal.c
 msgid "zi must be an ndarray"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/scipy/signal/signal.c
 msgid "zi must be of float type"
 msgstr ""
 
-#: extmod/ulab/code/filter/filter.c
+#: extmod/ulab/code/scipy/signal/signal.c
 msgid "zi must be of shape (n_section, 2)"
 msgstr ""
diff --git a/py/py.mk b/py/py.mk
index 3f94aa67a0..ae12e1976d 100644
--- a/py/py.mk
+++ b/py/py.mk
@@ -106,9 +106,9 @@ $(BUILD)/extmod/modbtree.o: CFLAGS += $(BTREE_DEFS)
 endif
 
 ifeq ($(CIRCUITPY_ULAB),1)
-SRC_MOD += $(patsubst $(TOP)/%,%,$(wildcard $(TOP)/extmod/ulab/code/*.c))
-SRC_MOD += $(patsubst $(TOP)/%,%,$(wildcard $(TOP)/extmod/ulab/code/*/*.c))
-CFLAGS_MOD += -DCIRCUITPY_ULAB=1 -DMODULE_ULAB_ENABLED=1
+ULAB_SRCS := $(shell find $(TOP)/extmod/ulab/code -type f -name "*.c")
+SRC_MOD += $(patsubst $(TOP)/%,%,$(ULAB_SRCS))
+CFLAGS_MOD += -DCIRCUITPY_ULAB=1 -DMODULE_ULAB_ENABLED=1 -iquote $(TOP)/extmod/ulab/code
 $(BUILD)/extmod/ulab/code/%.o: CFLAGS += -Wno-missing-declarations -Wno-missing-prototypes -Wno-unused-parameter -Wno-float-equal -Wno-sign-compare -Wno-cast-align -Wno-shadow -DCIRCUITPY
 endif
 
diff --git a/shared-bindings/_typing/__init__.pyi b/shared-bindings/_typing/__init__.pyi
index cc4a0a4391..9a6e963ea5 100644
--- a/shared-bindings/_typing/__init__.pyi
+++ b/shared-bindings/_typing/__init__.pyi
@@ -13,7 +13,7 @@ import rgbmatrix
 import ulab
 
 ReadableBuffer = Union[
-    bytes, bytearray, memoryview, array.array, ulab.array, rgbmatrix.RGBMatrix
+    bytes, bytearray, memoryview, array.array, ulab.ndarray, rgbmatrix.RGBMatrix
 ]
 """Classes that implement the readable buffer protocol
 
@@ -21,19 +21,19 @@ ReadableBuffer = Union[
   - `bytearray`
   - `memoryview`
   - `array.array`
-  - `ulab.array`
+  - `ulab.ndarray`
   - `rgbmatrix.RGBMatrix`
 """
 
 WriteableBuffer = Union[
-    bytearray, memoryview, array.array, ulab.array, rgbmatrix.RGBMatrix
+    bytearray, memoryview, array.array, ulab.ndarray, rgbmatrix.RGBMatrix
 ]
 """Classes that implement the writeable buffer protocol
 
   - `bytearray`
   - `memoryview`
   - `array.array`
-  - `ulab.array`
+  - `ulab.ndarray`
   - `rgbmatrix.RGBMatrix`
 """
 
diff --git a/shared-bindings/displayio/Bitmap.c b/shared-bindings/displayio/Bitmap.c
index ebe2946091..3b2a836eb9 100644
--- a/shared-bindings/displayio/Bitmap.c
+++ b/shared-bindings/displayio/Bitmap.c
@@ -43,7 +43,7 @@
 //| per row is a multiple of 4, then the resulting memoryview will correspond directly with the bitmap's contents. Otherwise,
 //| the bitmap data is packed into the memoryview with unspecified padding.
 //|
-//| A read-only buffer can be used e.g., with `ulab.frombuffer` to efficiently create an array with the same content as a Bitmap;
+//| A read-only buffer can be used e.g., with ``ulab.numpy.frombuffer`` to efficiently create an array with the same content as a Bitmap;
 //| to move data efficiently from ulab back into a Bitmap, use `bitmaptools.arrayblit`.
 //| """
 //|
diff --git a/shared-bindings/rgbmatrix/RGBMatrix.c b/shared-bindings/rgbmatrix/RGBMatrix.c
index 0188fa0b60..e71bae9f59 100644
--- a/shared-bindings/rgbmatrix/RGBMatrix.c
+++ b/shared-bindings/rgbmatrix/RGBMatrix.c
@@ -164,7 +164,7 @@ STATIC void preflight_pins_or_throw(uint8_t clock_pin, uint8_t *rgb_pins, uint8_
 //|         "RGB565" means that it is organized as a series of 16-bit numbers
 //|         where the highest 5 bits are interpreted as red, the next 6 as
 //|         green, and the final 5 as blue.  The object can be any buffer, but
-//|         `array.array` and `ulab.array` objects are most often useful.
+//|         `array.array` and ``ulab.ndarray`` objects are most often useful.
 //|         To update the content, modify the framebuffer and call refresh.
 //|
 //|         If a framebuffer is not passed in, one is allocated and initialized