Summary

Getting errors posting in the help forum so posting here. Having issues with Python runtime locally vs on Vercel. Would love some help/suggestions!

pip show openai is defined everywhere locally, but opening the fil in api/completions/utils/llm_response.py it gives me a Import "openai" could not be resolved error. When I run through vercel.dev with requirements.txt containing openai at my root, everything works fine on localhost, no issue.

However, in production on Vercel, the routes work so the api loads the functions correctly, but when the index.py file calls the llm_response.py file I'm getting the error below:
stdout: b''
stderr: b'Traceback (most recent call last):\n  File "/var/task/api/transcribe/whisper.py", line 1, in <module>\n    import openai\nModuleNotFoundError: No module named 'openai'\n'
stdout: b''
stderr: b'Traceback (most recent call last):\n  File "/var/task/api/transcribe/whisper.py", line 1, in <module>\n    import openai\nModuleNotFoundError: No module named 'openai'\n'

This is the api/completions/utils/llm_response.py file that's being called by api/completions/index.py.

import openai
import os

...

os.getenv("OPENAI_API_KEY")
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")

...

This is the repo I am working in:
https://github.com/SilenNaihin/roadman-talk

Example

https://roadman-talk.vercel.app/

Steps to Reproduce

git clone the repo I shared. then npm i for all the modules. Then vercel dev to run the Vercel cli - this works. Add your env variables at .env.local from the .env.template. Upload to Vercel and try to run any of the api calls