Skip to content

Fix workflow AI gating logic and add retries to Cerebrium preflight #69

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Rohan5commit merged 1 commit into from
May 23, 2026
Merged

Fix workflow AI gating logic and add retries to Cerebrium preflight #69

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
28 changes: 14 additions & 14 deletions .github/workflows/daily_trading_bot.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -196,15 +196,15 @@ jobs:

- name: Install Lightning dependencies
id: install_lightning_deps
if: ${{ steps.install_base_dependencies.outcome == 'success' && ((github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) || (github.event_name != 'workflow_dispatch') || (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading)) }}
if: ${{ steps.install_base_dependencies.outcome == 'success' && ((github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) || ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading)) }}
continue-on-error: ${{ !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) }}
timeout-minutes: 20
run: |
python -m pip install "lightning[app]==2.3.2" lightning-cloud==0.5.70

- name: Plan AI runtime
id: plan_ai_runtime
if: ${{ !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) }}
if: ${{ !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) }}
run: |
mkdir -p results
configured_inference_url="${CEREBRIUM_TRAINED_MODEL_URL}"
Expand Down Expand Up @@ -239,7 +239,7 @@ jobs:

- name: Validate Cerebrium primary configuration
id: validate_cerebrium_primary_config
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) }}
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) }}
run: |
runtime_mode="${{ steps.plan_ai_runtime.outputs.runtime_mode }}"
selected_backend="${{ steps.plan_ai_runtime.outputs.selected_backend }}"
Expand Down Expand Up @@ -276,7 +276,7 @@ jobs:

- name: Enforce AI routing invariants
id: enforce_ai_routing_invariants
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) }}
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) }}
run: |
runtime_mode="${{ steps.plan_ai_runtime.outputs.runtime_mode }}"
has_inference_url="${{ steps.validate_cerebrium_primary_config.outputs.has_inference_url }}"
Expand All @@ -294,7 +294,7 @@ jobs:
fi

- name: Emit AI runtime decision
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) }}
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) }}
run: |
echo "AI runtime mode: ${{ steps.plan_ai_runtime.outputs.runtime_mode }}"
echo "AI backend: ${{ steps.plan_ai_runtime.outputs.selected_backend }}"
Expand All @@ -303,7 +303,7 @@ jobs:

- name: Warm Cerebrium inference app
id: warm_cerebrium_inference
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && steps.enforce_ai_routing_invariants.outcome == 'success' && steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' }}
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && steps.enforce_ai_routing_invariants.outcome == 'success' && steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' }}
continue-on-error: true
timeout-minutes: 45
env:
Expand Down Expand Up @@ -367,7 +367,7 @@ jobs:

- name: Verify Cerebrium predict endpoint
id: verify_cerebrium_predict
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && steps.enforce_ai_routing_invariants.outcome == 'success' && steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' }}
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && steps.enforce_ai_routing_invariants.outcome == 'success' && steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' }}
continue-on-error: true
timeout-minutes: 3
env:
Expand Down Expand Up @@ -428,7 +428,7 @@ jobs:

- name: Run AI Trading Bot on Cerebrium
id: run_ai_bot_cerebrium
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && steps.enforce_ai_routing_invariants.outcome == 'success' && steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' && steps.verify_cerebrium_predict.outcome == 'success' }}
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && steps.enforce_ai_routing_invariants.outcome == 'success' && steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' && steps.verify_cerebrium_predict.outcome == 'success' }}
continue-on-error: true
timeout-minutes: 90
env:
Expand Down Expand Up @@ -459,7 +459,7 @@ jobs:

- name: Run AI Trading Bot on Cerebrium (Retry)
id: run_ai_bot_cerebrium_retry
if: ${{ always() && steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && steps.enforce_ai_routing_invariants.outcome == 'success' && steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' && steps.verify_cerebrium_predict.outcome == 'success' && steps.run_ai_bot_cerebrium.outcome == 'failure' }}
if: ${{ always() && steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && steps.enforce_ai_routing_invariants.outcome == 'success' && steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' && steps.verify_cerebrium_predict.outcome == 'success' && steps.run_ai_bot_cerebrium.outcome == 'failure' }}
continue-on-error: true
timeout-minutes: 90
env:
Expand Down Expand Up @@ -490,7 +490,7 @@ jobs:

- name: Launch Lightning inference studio
id: launch_lightning_inference
if: ${{ !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && steps.plan_ai_runtime.outputs.runtime_mode == 'lightning_full' && steps.install_lightning_deps.outcome == 'success' }}
if: ${{ !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && steps.plan_ai_runtime.outputs.runtime_mode == 'lightning_full' && steps.install_lightning_deps.outcome == 'success' }}
continue-on-error: true
timeout-minutes: 60
env:
Expand Down Expand Up @@ -519,7 +519,7 @@ jobs:

- name: Run AI Trading Bot in Lightning Studio
id: run_ai_bot_in_lightning_studio
if: ${{ !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && steps.plan_ai_runtime.outputs.runtime_mode == 'lightning_full' && steps.install_lightning_deps.outcome == 'success' && steps.launch_lightning_inference.outcome == 'success' }}
if: ${{ !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && steps.plan_ai_runtime.outputs.runtime_mode == 'lightning_full' && steps.install_lightning_deps.outcome == 'success' && steps.launch_lightning_inference.outcome == 'success' }}
continue-on-error: true
timeout-minutes: 90
env:
Expand Down Expand Up @@ -553,7 +553,7 @@ jobs:

- name: Run AI Trading Bot (Distilled Local Fallback)
id: run_ai_bot_distilled_local
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && (steps.plan_ai_runtime.outputs.runtime_mode == 'distilled_local' || (steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' && (steps.warm_cerebrium_inference.outcome == 'failure' || steps.verify_cerebrium_predict.outcome == 'failure' || (steps.run_ai_bot_cerebrium.outcome == 'failure' && steps.run_ai_bot_cerebrium_retry.outcome != 'success'))) || (steps.plan_ai_runtime.outputs.runtime_mode == 'lightning_full' && (steps.launch_lightning_inference.outcome == 'failure' || steps.run_ai_bot_in_lightning_studio.outcome == 'failure'))) }}
if: ${{ steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && (steps.plan_ai_runtime.outputs.runtime_mode == 'distilled_local' || (steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true' && (steps.warm_cerebrium_inference.outcome == 'failure' || steps.verify_cerebrium_predict.outcome == 'failure' || (steps.run_ai_bot_cerebrium.outcome == 'failure' && steps.run_ai_bot_cerebrium_retry.outcome != 'success'))) || (steps.plan_ai_runtime.outputs.runtime_mode == 'lightning_full' && (steps.launch_lightning_inference.outcome == 'failure' || steps.run_ai_bot_in_lightning_studio.outcome == 'failure'))) }}
continue-on-error: true
timeout-minutes: 60
env:
Expand All @@ -575,7 +575,7 @@ jobs:

- name: Run AI Trading Bot (Emergency Distilled Retry)
id: run_ai_bot_distilled_emergency_retry
if: ${{ always() && steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && steps.run_ai_bot_distilled_local.outcome == 'failure' }}
if: ${{ always() && steps.install_base_dependencies.outcome == 'success' && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && steps.run_ai_bot_distilled_local.outcome == 'failure' }}
continue-on-error: true
timeout-minutes: 60
env:
Expand All @@ -599,7 +599,7 @@ jobs:

- name: Send AI Failure Report
id: send_ai_failure_report
if: ${{ always() && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && (github.event_name == 'workflow_dispatch' && !inputs.disable_ai_trading) && (((steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true') && (steps.warm_cerebrium_inference.outcome == 'failure' || steps.verify_cerebrium_predict.outcome == 'failure' || (steps.run_ai_bot_cerebrium.outcome == 'failure' && steps.run_ai_bot_cerebrium_retry.outcome != 'success'))) || ((steps.plan_ai_runtime.outputs.runtime_mode == 'lightning_full') && (steps.launch_lightning_inference.outcome == 'failure' || steps.run_ai_bot_in_lightning_studio.outcome == 'failure') && steps.run_ai_bot_distilled_local.outcome == 'failure' && steps.run_ai_bot_distilled_emergency_retry.outcome != 'success') || ((steps.plan_ai_runtime.outputs.runtime_mode == 'distilled_local') && steps.run_ai_bot_distilled_local.outcome == 'failure' && steps.run_ai_bot_distilled_emergency_retry.outcome != 'success')) }}
if: ${{ always() && !(github.event_name == 'workflow_dispatch' && inputs.run_in_lightning_studio) && ((github.event_name != 'workflow_dispatch') || !inputs.disable_ai_trading) && (((steps.plan_ai_runtime.outputs.runtime_mode == 'cerebrium_full' && steps.validate_cerebrium_primary_config.outputs.has_inference_url == 'true') && (steps.warm_cerebrium_inference.outcome == 'failure' || steps.verify_cerebrium_predict.outcome == 'failure' || (steps.run_ai_bot_cerebrium.outcome == 'failure' && steps.run_ai_bot_cerebrium_retry.outcome != 'success'))) || ((steps.plan_ai_runtime.outputs.runtime_mode == 'lightning_full') && (steps.launch_lightning_inference.outcome == 'failure' || steps.run_ai_bot_in_lightning_studio.outcome == 'failure') && steps.run_ai_bot_distilled_local.outcome == 'failure' && steps.run_ai_bot_distilled_emergency_retry.outcome != 'success') || ((steps.plan_ai_runtime.outputs.runtime_mode == 'distilled_local') && steps.run_ai_bot_distilled_local.outcome == 'failure' && steps.run_ai_bot_distilled_emergency_retry.outcome != 'success')) }}
continue-on-error: true
env:
SMTP_SERVER: ${{ secrets.SMTP_SERVER }}
Expand Down
28 changes: 21 additions & 7 deletions run_ai_daily_cerebrium.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,7 @@
import os
import subprocess
import sys
import time
from pathlib import Path

import requests
Expand Down Expand Up @@ -41,13 +42,26 @@ def _preflight_predict(url: str, api_key: str) -> tuple[bool, str]:
if api_key:
headers["Authorization"] = f"Bearer {api_key}"
payload = {"candidates": [{"symbol": "AAPL", "return_5d": 0.01, "news_sentiment_7d": 0.0}]}
try:
resp = requests.post(url, headers=headers, json=payload, timeout=45)
if 200 <= resp.status_code < 300:
return True, f"status={resp.status_code}"
return False, f"status={resp.status_code}"
except Exception as exc:
return False, str(exc)

attempts = 3
transient_statuses = {429, 500, 502, 503, 504}
details: list[str] = []

for attempt in range(1, attempts + 1):
try:
resp = requests.post(url, headers=headers, json=payload, timeout=45)
details.append(f"attempt={attempt}:status={resp.status_code}")
if 200 <= resp.status_code < 300:
return True, "; ".join(details)
if resp.status_code not in transient_statuses:
return False, "; ".join(details)
except requests.RequestException as exc:
details.append(f"attempt={attempt}:exception={exc}")

if attempt < attempts:
time.sleep(min(8, 2 * attempt))

return False, "; ".join(details)


def main() -> None:
Expand Down