This repository contains a development environment that is provided as an easy starting point for Hackathon participants to use to get off and running.
Use of this dev environment is not required but is encouraged to allow Hackathon staff to be able to easily help you when troubleshooting issues. It also offers great standard rapid prototyping tools that will accelerate your development while not having to think about fine details like how to build a UI or API backend. You can also be assured that your teammates will all have the same environment and spend less time troubleshooting, and more time hacking!
That said, if you truly are experts, don't let us hold you back! You are free to use whatever technologies and development environments that you want to use in this hackathon!
Using this development environment has a few prerequisites:
- Install Docker Desktop on your computer (no need to create an account)
- Install VS Code on your computer
- Install the "Dev Containers" extension inside VS Code
- Install the AWS CLI V2 on your computer (v2 is required, v1 will not work)
- Open VS Code, clone this repository to your local machine and open it
- Before doing anything create a .env file in the root of the repo with any secrets provided to you by staff (e.g. OPENAI_API_KEY, OPENAI_BASE_URL, etc.)
# .env file
OPENAI_API_KEY=https://api.ai.it.cornell.edu
OPENAI_BASE_URL=sk-abcdefg12-3
- In the bottom left corner of VS Code, click the green button and select "Reopen in Container" from the menu
- Wait for some time while the Dev Container environment is built on your machine
- Once the process is complete, you should see a file tree and open a terminal within the Dev Container workspace!
⛔️ if you get an error like the following, you likely did not create a .env file in step 2. You will need to delete the .env directories that got created under the following directories:
Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error mounting "/host_mnt/Users/mjs472/repos/ai-hackathon-2025/.env" to rootfs at "/code/.env": mount /host_mnt/Users/mjs472/repos/ai-hackathon-2025/.env:/code/.env (via /proc/self/fd/6), flags: 0x5000: not a directory: unknown: Are you trying to mount a directory onto a file (or vice-versa)? Check if the specified host path exists and is the expected type
- the repo root directory (./.env)
- the streamlit-app directory (./streamlit-app/.env)
- the fastapi-app directory (./fastapi-app/.env)
Once you delete these, try to "Reopen in Container" again.
Under the notebooks/ directory, you can create python notebooks. Simply create a new file named <filename>.ipynb and you can open it from the file navigator. This will allow you to edit and run code in the notebook right inside VS Code!
Python 3.12.8
Streamlit is a powerful web UI prototyping tool that works in pure python. You can create chatbots, forms with a rich set of inputs, complex layouts, etc. all that work within a web browser!
The dev environment has a separate container running the streamlit app in the background. It will be running on port 8501 on your local machine so you can open your browser and navigate to http://localhost:8501/ to test it out.
The code for the Streamlit app is under streamlit-app/app. Any updates to the streamlit-app/app/Home.py or adjacent files will be automatically detected and reload the streamlit server, so you can easily make changes and test them live!
FastAPI is a powerful asynchronous API building library. You can build a custom API for your application with this!
Documentation: https://fastapi.tiangolo.com/
The dev environment has a separate container running the FastAPI server in the background. It will be running on port 8000 on your local machine so you can open your browser and navigate to http://localhost:8000/docs to see the available endpoints and to test them out! You can also use something like curl from your local machine too to call the api endpoints via CLI, or obviously, you can call them with something like the requests module in python!
The code for the FastAPI app is under fastapi-app/app. Any updates to the fastapi-app/app/server.py or adjacent files will be automatically detected and reload the FastAPI server, so you can easily make changes and test them live!
localhost. e.g. https://fastapi:8000/my-api-endpoint
Neo4j is a graph database that allows us to store and query data in a graph format. This can be useful for storing data that has relationships between entities, such as social networks, recommendation systems, and knowledge graphs.
Documentation: https://neo4j.com/docs/
The dev environment has a separate container running the Neo4j server in the background. The graph db server that you can connect to from code is running on port 7687. The db web interface that you can access via your browser is running on port 7474. You can navigate to http://localhost:7474 and log in!
Username: neo4j
Password: secretpassword
localhost. e.g. neo4j://neo4j:7687
Many common frameworks and libraries are already installed in the dev environment for your use. These include things like:
- Pydantic
- LangChain
- LangGraph
- Llama-Index
- ChromaDB (for in-memory vector db)
- Neo4j (for graph db, see above for how to connect)
- PyTorch & Ray
- Matplotlib, Plotly, Seaborn (for visualization)
- Pandas
- GeoPandas, CartoPy, GDAL (for maps / geospatial data)
- Huggingface Transformers and Diffusers
- NumPy, SciPy, Dask
- Scikit-learn
- FFMPEG
- OpenCV-Python
- Mediapipe
- xgboost, nltk (and other various ML-related modules)
- Others, check out
pyproject.tomlfor the full list
If you are missing a Python module you would like to use, it can be added as a dependency in the pyproject.toml file, running poetry lock, and rebuilding the Dev Container.
To add native libraries, the Dockerfile can be updated to install via the apt package manager and rebuilding the Dev Container. You could also update the Dockerfile to download, build and install something from source in a pinch.
Feel free to ask hackathon staff for assistance.
The thisapp directory is set up so that all running components will have access to the python code underneath as a module.
For example, if you wanted to create a FastAPI backend and Streamlit frontend that both use a common Pydantic model named MyModel, you could define it under the thisapp directory in a python file named thisapp/my_model.py and then both apps can simply do from thisapp.my_model import MyModel and have access to the same code!