diff --git a/crates/service/job/inference_job/src/job/job_types/format_conversion/fbx_to_gltf/process_job.rs b/crates/service/job/inference_job/src/job/job_types/format_conversion/fbx_to_gltf/process_job.rs
index e0adf45aa9..cbd8afe828 100644
--- a/crates/service/job/inference_job/src/job/job_types/format_conversion/fbx_to_gltf/process_job.rs
+++ b/crates/service/job/inference_job/src/job/job_types/format_conversion/fbx_to_gltf/process_job.rs
@@ -28,6 +28,7 @@ use crate::job_dependencies::JobDependencies;
 use crate::util::maybe_download_file_from_bucket::{maybe_download_file_from_bucket, MaybeDownloadArgs};
 
 const BUCKET_FILE_EXTENSION : &str = ".gltf";
+const BUFFER_BIN_FILE_EXTENSION : &str = ".bin";
 
 pub struct FbxToGltfJobArgs<'a> {
   pub job_dependencies: &'a JobDependencies,
@@ -152,62 +153,88 @@ pub async fn process_job(args: FbxToGltfJobArgs<'_>) -> Result<JobSuccessResult,
 
   // ==================== CHECK ALL FILES EXIST AND GET METADATA ==================== //
 
-  // NB: The actual name of the file will be this:
-  let output_gltf_file = output_directory_actual.join("output.gltf"); // NB: FBX2GLTF creates multiple files in the directory
+  // NB: FBX2GLTF creates multiple files in the directory
+  // NB: The actual name of the file will be these:
+  let output_gltf_file = output_directory_actual.join("output.gltf");
+  let output_buffer_file = output_directory_actual.join("buffer.bin");
 
   info!("Checking that output files exist...");
 
   check_file_exists(&output_gltf_file).map_err(|e| ProcessSingleJobError::Other(e))?;
+  check_file_exists(&output_buffer_file).map_err(|e| ProcessSingleJobError::Other(e))?;
 
   info!("Interrogating result file properties...");
 
-  let file_size_bytes = file_size(&output_gltf_file)
+  let gltf_file_size_bytes = file_size(&output_gltf_file)
       .map_err(|err| ProcessSingleJobError::Other(err))?;
 
-  let maybe_mimetype = get_mimetype_for_file(&output_gltf_file)
+  let maybe_gltf_mimetype = get_mimetype_for_file(&output_gltf_file)
       .map_err(|err| ProcessSingleJobError::from_io_error(err))?
       .map(|mime| mime.to_string());
 
   info!("Calculating sha256...");
 
-  let file_checksum = sha256_hash_file(&output_gltf_file)
+  let gltf_file_checksum = sha256_hash_file(&output_gltf_file)
       .map_err(|err| {
         ProcessSingleJobError::Other(anyhow!("Error hashing file: {:?}", err))
       })?;
 
-  // ==================== UPLOAD AUDIO TO BUCKET ==================== //
+  // ==================== UPLOAD GLTF TO BUCKET ==================== //
 
-  job_progress_reporter.log_status("uploading result")
+  job_progress_reporter.log_status("uploading output.gltf")
       .map_err(|e| ProcessSingleJobError::Other(e))?;
 
-  let result_bucket_location = MediaFileBucketPath::generate_new(
+  let result_gtlf_bucket_location = MediaFileBucketPath::generate_new(
     None,
     Some(BUCKET_FILE_EXTENSION));
 
-  let result_bucket_object_pathbuf = result_bucket_location.to_full_object_pathbuf();
+  let result_gltf_bucket_object_pathbuf = result_gtlf_bucket_location.to_full_object_pathbuf();
 
-  info!("Destination bucket path: {:?}", &result_bucket_object_pathbuf);
+  info!("Destination bucket path: {:?}", &result_gltf_bucket_object_pathbuf);
 
-  info!("Uploading media file...");
+  info!("Uploading...");
 
   args.job_dependencies.buckets.public_bucket_client.upload_filename_with_content_type(
-    &result_bucket_object_pathbuf,
+    &result_gltf_bucket_object_pathbuf,
     &output_gltf_file,
-    "audio/wav")
+    "model/gltf+json")
       .await
       .map_err(|e| ProcessSingleJobError::Other(e))?;
 
+  // ==================== UPLOAD BIN TO BUCKET ==================== //
+
+  job_progress_reporter.log_status("uploading buffer.bin")
+      .map_err(|e| ProcessSingleJobError::Other(e))?;
+
+  // NB: Reuse same object hash, just change the extension.
+  let result_buffer_bucket_location = MediaFileBucketPath::from_object_hash(
+    result_gtlf_bucket_location.get_object_hash(),
+    None,
+    Some(BUFFER_BIN_FILE_EXTENSION));
+
+  let result_buffer_bucket_object_pathbuf = result_buffer_bucket_location.to_full_object_pathbuf();
+
+  info!("Destination bucket path: {:?}", &result_buffer_bucket_object_pathbuf);
+
+  info!("Uploading...");
+
+  args.job_dependencies.buckets.public_bucket_client.upload_filename_with_content_type(
+    &result_buffer_bucket_object_pathbuf,
+    &output_buffer_file,
+    "application/octet-stream")
+      .await
+      .map_err(|e| ProcessSingleJobError::Other(e))?;
+
+  // ==================== CLEANUP FILES ==================== //
+
+  safe_delete_temp_file(&output_buffer_file);
   safe_delete_temp_file(&output_gltf_file);
   safe_delete_temp_directory(&output_directory_actual);
-
-  // ==================== DELETE DOWNLOADED FILE ==================== //
-
-  // NB: We should be using a tempdir, but to make absolutely certain we don't overflow the disk...
   safe_delete_temp_directory(&work_temp_dir);
 
   // ==================== SAVE RECORDS ==================== //
 
-  info!("Saving record...");
+  info!("Saving record (only gltf, not the buffer.bin) ...");
 
   let (inference_result_token, id) = insert_media_file_generic(InsertArgs {
     pool: &args.job_dependencies.db.mysql_pool,
@@ -218,30 +245,30 @@ pub async fn process_job(args: FbxToGltfJobArgs<'_>) -> Result<JobSuccessResult,
     maybe_origin_model_type: None,
     maybe_origin_model_token: None,
     maybe_origin_filename: None,
-    maybe_mime_type: maybe_mimetype.as_deref(),
-    file_size_bytes,
+    maybe_mime_type: maybe_gltf_mimetype.as_deref(),
+    file_size_bytes: gltf_file_size_bytes,
     maybe_duration_millis: None,
     maybe_audio_encoding: None,
     maybe_video_encoding: None,
     maybe_frame_width: None,
     maybe_frame_height: None,
-    public_bucket_directory_hash: result_bucket_location.get_object_hash(),
-    maybe_public_bucket_prefix: result_bucket_location.get_optional_prefix(),
-    maybe_public_bucket_extension: result_bucket_location.get_optional_extension(),
+    public_bucket_directory_hash: result_gtlf_bucket_location.get_object_hash(),
+    maybe_public_bucket_prefix: result_gtlf_bucket_location.get_optional_prefix(),
+    maybe_public_bucket_extension: result_gtlf_bucket_location.get_optional_extension(),
     extra_file_modification_info: None,
     maybe_creator_file_synthetic_id_category: IdCategory::MediaFile,
     maybe_creator_category_synthetic_id_category: IdCategory::MocapResult,
     maybe_mod_user_token: None,
     is_batch_generated: false,
     is_generated_on_prem: false,
-    checksum_sha2: &file_checksum,
+    checksum_sha2: &gltf_file_checksum,
     generated_by_worker: Some(&args.job_dependencies.job.info.container.hostname),
     generated_by_cluster: Some(&args.job_dependencies.job.info.container.cluster_name),
   })
       .await
       .map_err(|e| ProcessSingleJobError::Other(e))?;
 
-  info!("VC Done.");
+  info!("Fbx2Gltf Done.");
 
   //args.job_dependencies.clients.firehose_publisher.vc_inference_finished(
   //  maybe_user_token.as_ref(),