diff --git a/Cargo.lock b/Cargo.lock
index b9d40aa954..149424ab0a 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2628,6 +2628,7 @@ dependencies = [
 name = "fakeyou_client"
 version = "0.0.1"
 dependencies = [
+ "backoff",
  "bytes 1.3.0",
  "chrono",
  "enums",
diff --git a/crates/lib/fakeyou_client/Cargo.toml b/crates/lib/fakeyou_client/Cargo.toml
index 9f79f68198..2d33b2a31d 100644
--- a/crates/lib/fakeyou_client/Cargo.toml
+++ b/crates/lib/fakeyou_client/Cargo.toml
@@ -31,6 +31,7 @@ serde_json = "1.0.64"
 toml = "0.5.8"
 
 # HTTP
+backoff = { version = "0.4.0", features = ["futures"] } # Exponential backoff retry
 reqwest = { version = "0.11", features = ["json", "cookies"] } # HTTP Client
 
 [dev-dependencies]
diff --git a/crates/lib/fakeyou_client/src/fakeyou_api_client.rs b/crates/lib/fakeyou_client/src/fakeyou_api_client.rs
index 7344e99fa3..9dc8c9ed8a 100644
--- a/crates/lib/fakeyou_client/src/fakeyou_api_client.rs
+++ b/crates/lib/fakeyou_client/src/fakeyou_api_client.rs
@@ -1,9 +1,12 @@
-use std::sync::Arc;
+use backoff::future::retry;
+use backoff::{ExponentialBackoff, ExponentialBackoffBuilder};
 use crate::api::tts_inference::{CreateTtsInferenceRequest, CreateTtsInferenceResponse, TtsInferenceJobStatus};
-use errors::AnyhowResult;
-use reqwest::{Client, ClientBuilder, RequestBuilder, Url};
-use reqwest::cookie::Jar;
 use crate::credentials::FakeYouCredentials;
+use errors::{anyhow, AnyhowResult};
+use reqwest::cookie::Jar;
+use reqwest::{Client, ClientBuilder, RequestBuilder, Url};
+use std::sync::Arc;
+use std::time::Duration;
 
 const AUTHORIZATION_HEADER: &'static str = "Authorization";
 const SESSION_COOKIE_NAME : &'static str = "session";
@@ -55,18 +58,37 @@ impl FakeYouApiClient {
 
   // TODO: need to yield better, more "library"-appropriate errors.
   pub async fn create_tts_inference(&self, request: CreateTtsInferenceRequest<'_>) -> AnyhowResult<CreateTtsInferenceResponse> {
-    let url = format!("https://{}/tts/inference", self.api_domain);
+    let response = self.do_create_tts_inference(&request).await?;
+    let response = serde_json::from_str(&response)?;
+    Ok(response)
+  }
 
+  pub async fn create_tts_inference_with_backoff(&self, request: CreateTtsInferenceRequest<'_>) -> AnyhowResult<CreateTtsInferenceResponse> {
+    //let backoff = ExponentialBackoff::default();
+    let mut backoff = ExponentialBackoffBuilder::new()
+        .with_initial_interval(Duration::from_secs(3)) // Initial retry interval: 3 seconds
+        .with_randomization_factor(0.5) // 50% below, 50% above interval
+        .with_multiplier(2.0) // Double the wait
+        .with_max_elapsed_time(Some(Duration::from_secs(60)))
+        //.with_max_interval(Duration::from_secs(30))
+        .build();
+    let response = retry(backoff, || async {
+      Ok(self.do_create_tts_inference(&request).await?)
+    }).await?;
+
+    let response = serde_json::from_str(&response)?;
+    Ok(response)
+  }
+
+  pub async fn do_create_tts_inference(&self, request: &CreateTtsInferenceRequest<'_>) -> Result<String, reqwest::Error> {
+    let url = format!("https://{}/tts/inference", self.api_domain);
     let response = self.add_credentials(self.client
         .post(url))
-        .json(&request)
+        .json(request)
         .send()
         .await?
         .text()
         .await?;
-
-    let response = serde_json::from_str(&response)?;
-
     Ok(response)
   }
 
diff --git a/crates/programs/aichatbot-sidecar/src/workers/audio/fakeyou_audio_create/process_single_record.rs b/crates/programs/aichatbot-sidecar/src/workers/audio/fakeyou_audio_create/process_single_record.rs
index 6ee5dd39ba..0eb28aca28 100644
--- a/crates/programs/aichatbot-sidecar/src/workers/audio/fakeyou_audio_create/process_single_record.rs
+++ b/crates/programs/aichatbot-sidecar/src/workers/audio/fakeyou_audio_create/process_single_record.rs
@@ -21,7 +21,7 @@ pub async fn process_single_record(target: &TtsRenderTask, job_state: &Arc<JobSt
 
   let result = job_state
       .fakeyou_client
-      .create_tts_inference(CreateTtsInferenceRequest {
+      .create_tts_inference_with_backoff(CreateTtsInferenceRequest {
         uuid_idempotency_token: &uuid_idempotency_token,
         tts_model_token: &tts_model_token,
         inference_text: &target.full_text,
@@ -39,7 +39,7 @@ pub async fn process_single_record(target: &TtsRenderTask, job_state: &Arc<JobSt
     },
     Ok(res) => {
       if !res.success {
-        error!("unknown problem in submitting to FakeYou");
+        error!("unknown problem in submitting TTS request to FakeYou");
 
         update_tts_render_task_unsuccessfully_submitted(UnsuccessfulArgs {
           tts_render_task_token: &target.token,
diff --git a/crates/programs/aichatbot-sidecar/src/workers/audio/fakeyou_audio_download/process_single_record.rs b/crates/programs/aichatbot-sidecar/src/workers/audio/fakeyou_audio_download/process_single_record.rs
index 7a9a67827e..f573ac505d 100644
--- a/crates/programs/aichatbot-sidecar/src/workers/audio/fakeyou_audio_download/process_single_record.rs
+++ b/crates/programs/aichatbot-sidecar/src/workers/audio/fakeyou_audio_download/process_single_record.rs
@@ -40,7 +40,7 @@ pub async fn process_single_record(target: &TtsRenderTask, job_state: &Arc<JobSt
     Ok(res) => {
       match res.state {
         None => {
-          error!("unknown problem in submitting to FakeYou");
+          error!("unknown problem in checking FakeYou TTS inference job status");
           // TODO: Errors
           //update_tts_render_task_unsuccessfully_submitted(UnsuccessfulArgs {
           //  tts_render_task_token: &target.token,