BackendStarterProject

Note - this task as is, is just a skeleton which closely resembles the actual starter task once the long term hiring decision is made. The long term hiring decision will be made based on the time estimate + actual delivery time + code quality + quality of work on this skeleton task.

Problem statement

Create a fastapi + gcp workflow + bigquery + gcp firebase stack that satisfies the requirements.

Details on each step

---

Fastapi service 1

Fast api service needs to have 2 primary endpoints(and more auxiliary helper end points as you prefer) -

GET endpoint - is_valid_input(x) : which returns the string “Success” if the input is a string all in lower case, “Fail” is the input is a string that has upper case, returns a 4XX error with a message (“{data_type}” is not allowed) if the data type is an int or float and a 5XX( run time exception if the input is a Boolean)

Post end point - process(x) which should sleep for 3 mins, write x as a string to a bigquery table(that has just one field with the name input) before returning “Success”

Server definition for FastAPI service 1 can be found under fastapi-server1 directory

---

Gcp workflow

(Link - https://cloud.google.com/workflows)

The workflow should do the following

1. Accept an input called name

2. Call the is_valid_input argument with the input name.

   1. If the resultant http error code is 429, retry with exponential back off upto 3 times

   2. If the result is a 4XX/5XX, or a 2XX with message as “Fail” go directly to step 6 and write the error message onto bigquery

   3. Update to fire base with the key as x and value as {“done”:[“valid_input_check”]}

3. Call process with x

4. Update to firebase with the document id x and append [“process”] to the key “done”

5. Load the result of valid_input_check and process to big query into a table which has 4 columns (name, valid_input_check, process, timestamp)

6. Return (name, valid_input_check, process) as output from workflow

Workflow definition can be found in WorkFlow.yaml

---

Fast api service 2

It has 2 endpoints -

trigger workflow(x) which should trigger the workflow and return the workflow id

check status(x) which should

1. Check if workflow is done by checking the workflow status api - https://cloud.google.com/workflows/docs/reference/rest/v1beta/projects.locations.workflows/get
2. If the workflow is not done, check the state from firebase and return the state.

Server definition for FastAPI service 2 can be found under fastapi-server2 directory

---

Firebase

Use a firestore (https://firebase.google.com/products/firestore) to have a simple key value pair that will be used to update state in the workflow and check state in fast api service 2.

---

Task instructions

Use your own cloud for this task(most of these would be free for a small scale).

Send the GitHub link to check the code once approved to start the task(if the estimate sounds reasonable)

---

Deployment

1. Clone the repo in a server with: git clone https://github.com/subhayukumar/BackendStarterProject
2. Run cd BackendStarterProject/fastapi-server1 && pip install -r requirements.txt && cd - to install the dependencies for API server 1.
3. Run cd BackendStarterProject/fastapi-server2 && pip install -r requirements.txt && cd - to install the dependencies for API server 2.
4. Get a JSON key file for authenticating to Google Cloud and then run export GOOGLE_APPLICATION_CREDENTIALS="<path to JSON key file>"
5. Run cd fastapi-server1 && uvicorn main:app --reload --host 0.0.0.0 --port <PORT_1> in one terminal.
6. Run cd fastapi-server2 && uvicorn main:app --reload --host 0.0.0.0 --port <PORT_2> in another terminal.
7. Now visit (http://<PUBLIC_IP>:<PORT_2>/docs) for viewing available routes.# BackendStarterProject