diff --git a/pkgs/development/python-modules/tensorflow/bin.nix b/pkgs/development/python-modules/tensorflow/bin.nix
index 140a77942dab..5b816bee894c 100644
--- a/pkgs/development/python-modules/tensorflow/bin.nix
+++ b/pkgs/development/python-modules/tensorflow/bin.nix
@@ -20,6 +20,7 @@
   distutils,
   wheel,
   jax,
+  ml-dtypes,
   opt-einsum,
   tensorflow-estimator-bin,
   tensorboard,
@@ -38,9 +39,10 @@
   typing-extensions,
 }:
 
-# We keep this binary build for two reasons:
+# We keep this binary build for three reasons:
 # - the source build doesn't work on Darwin.
 # - the source build is currently brittle and not easy to maintain
+# - the source build doesn't work on NVIDIA Jetson platforms
 
 # unsupported combination
 assert !(stdenv.isDarwin && cudaSupport);
@@ -49,19 +51,19 @@ let
   packages = import ./binary-hashes.nix;
   inherit (cudaPackages) cudatoolkit cudnn;
 in
-buildPythonPackage {
+buildPythonPackage rec {
   pname = "tensorflow" + lib.optionalString cudaSupport "-gpu";
-  inherit (packages) version;
+  version = packages."${"version" + lib.optionalString (cudaSupport && cudaPackages.cudaFlags.isJetsonBuild) "_jetson"}";
   format = "wheel";
 
   src =
     let
       pyVerNoDot = lib.strings.stringAsChars (x: lib.optionalString (x != ".") x) python.pythonVersion;
       platform = stdenv.system;
-      cuda = lib.optionalString cudaSupport "_gpu";
+      cuda = lib.optionalString cudaSupport (if cudaPackages.cudaFlags.isJetsonBuild then "_jetson" else "_gpu");
       key = "${platform}_${pyVerNoDot}${cuda}";
     in
-    fetchurl (packages.${key} or (throw "tensoflow-bin: unsupported system: ${stdenv.system}"));
+    fetchurl (packages.${key} or (throw "tensoflow-bin: unsupported configuration: ${key}"));
 
   buildInputs = [ llvmPackages.openmp ];
 
@@ -73,7 +75,7 @@ buildPythonPackage {
     protobuf
     numpy
     scipy
-    jax
+    (if !cudaPackages.cudaFlags.isJetsonBuild then jax else ml-dtypes)
     termcolor
     grpcio
     six
@@ -103,6 +105,10 @@ buildPythonPackage {
 
     pushd dist
 
+    for f in tensorflow-*+nv*.whl; do
+      mv "$f" "$(sed -E 's/(nv[0-9]+)\.0*([0-9]+)/\1.\2/' <<< "$f")"
+    done
+
     wheel unpack --dest unpacked ./*.whl
     rm ./*.whl
     (
@@ -134,11 +140,12 @@ buildPythonPackage {
   postFixup =
     let
       # rpaths we only need to add if CUDA is enabled.
-      cudapaths = lib.optionals cudaSupport [
+      cudapaths = lib.optionals cudaSupport ([
         cudatoolkit.out
         cudatoolkit.lib
+      ] ++ lib.optionals (!cudaPackages.cudaFlags.isJetsonBuild) [
         cudnn
-      ];
+      ]);
 
       libpaths = [
         stdenv.cc.cc.lib