diff --git a/R/upload_cloud.R b/R/upload_cloud.R
new file mode 100644
index 0000000..7f3768a
--- /dev/null
+++ b/R/upload_cloud.R
@@ -0,0 +1,30 @@
+# how do we want to structure meta and counts container on the cloud?
+# how does file_id_db get generated? if load all counts into one container, there would be an issue if file_id_db in new atlas duplicate with that on the cloud 
+# Ideally a container should contain meta, and directories for raw counts and cpm
+
+assay_map <- c(counts = "original",
+               cpm = "cpm")
+
+testing <- read_parquet("testing.0.2.3.parquet")
+testing<- tibble(testing)
+
+upload_cloud <- function(meta,
+                         assay = "counts",
+                         version = "0.2.3",
+                         cache_dir = CuratedAtlasQueryR:::get_default_cache_dir()) {
+  #check if the software is in the directory
+  check_file_exists(file.path(cache_dir, "harmonised_human_cell_atlas-openrc.sh")) 
+  meta_input <- paste(meta, version, sep=".")
+  upload_meta <- glue("swift upload {meta_input} {cache_dir}/{meta_input}.parquet)")
+  upload_counts <- glue("swift upload {meta_input} {cache_dir}/{assay})")
+  # Swift software can be installed when python>=3.11
+  system("module load python/3.11", intern = TRUE,)
+  check_swift <- system("swift list --help", intern = TRUE, ignore.stderr = TRUE)
+  
+  if (length(check_swift) == 0){
+    system("python - pip install python-swiftclient", intern = TRUE)
+  }
+  system("source harmonised_human_cell_atlas-openrc.sh", intern = TRUE)
+  system(upload_meta)
+  system(upload_counts)
+}