Hackathon instructions

tutorials/hackathon_setup.md

@@ -0,0 +1,82 @@
+
+# 🐳 Minimal Jupyter via SSH + Docker (Lambda Cloud)
+
+This guide sets up Jupyter running **inside a Docker container** on your Lambda Cloud instance and forwards it securely to your local machine — without using Lambda's built-in JupyterLab.
+
+---
+
+## ✅ 1. SSH into your Lambda instance
+
+On your local machine:
+
+```bash
+ssh -i <YOUR_SSH_KEY_PATH> ubuntu@<INSTANCE_IP>
+```
+
+Replace `<YOUR_SSH_KEY_PATH>` with your private key path and `<INSTANCE_IP>` with your instance's IP address.
+
+---
+
+## ✅ 2. Start your Docker container with Jupyter
+
+```bash
+sudo docker run --gpus all --shm-size=64g -dit \
+  --name bionemo \
+  -p 8888:8888 \
+  -v /home/ubuntu/bionemo_workspace:/workspace \
+  nvcr.io/nvidia/clara/bionemo-framework:nightly
+```
+
+Enter the container:
+
+```bash
+sudo docker exec -it bionemo bash
+```
+
+---
+
+## ✅ 3. Install Jupyter and other Python tools inside the container
+
+Inside the container shell:
+
+```bash
+pip install jupyter rapids-singlecell
+git clone -b tahoe-DGX-fix https://github.com/theislab/vevo_Tahoe_100m_analysis.git
+```
+
+Then launch Jupyter:
+
+```bash
+jupyter lab --ip=0.0.0.0 --port=8888 --no-browser --allow-root
+```
+
+Note the access token printed in the terminal output (you’ll need it to log in).
+
+---
+
+## ✅ 4. Forward Jupyter port from remote to local
+
+On your **local machine**, open a new terminal and run:
+
+```bash
+ssh -i <YOUR_SSH_KEY_PATH> -L 8888:127.0.0.1:8888 ubuntu@<INSTANCE_IP>
+```
+
+Now visit:
+
+```
+http://localhost:8888/?token=<TOKEN>
+```
+
+Paste in the token you copied from the container output.
+
+---
+
+## ✅ Optional cleanup
+
+To stop and remove the container:
+
+```bash
+sudo docker stop bionemo
+sudo docker rm bionemo
+```