Python/Full Stack Developer for Interactive Dashboard
Budget: £250 – £750 GBP
We are seeking a skilled Full Stack developer (Python, React, SQL, Docker) to build an application that interacts with a locally installed Large Language Model (LLM).
We are looking for an Application that essentially:
- Loops through a directory of XML/json files
- Forms prompts using the contents of these files
- Systematically submits these prompts to a locally installed LLM
- Potentially use Redis for queueing prompts
- Saves the results of the prompts to a local PostgreSQL instance
- Also allows for the option to save to CSV (both options configurable as on/off)
- Build a basic React front-end to view and interact with the output (original data passed to LLM and the resulting information)
- The UI will eventually grown in complexity so should be open for extending/scalable
All of this infrastructure (including the LLM) should be included in the Docker configuration.
Should be built with eventual scale in mind and flexibility to run 1000s of documents through at a later stage.
Flexibility to swap out the LLM being used and even store the different results that different installed LLMs have returned.
We are looking for an Application that essentially:
- Loops through a directory of XML/json files
- Forms prompts using the contents of these files
- Systematically submits these prompts to a locally installed LLM
- Potentially use Redis for queueing prompts
- Saves the results of the prompts to a local PostgreSQL instance
- Also allows for the option to save to CSV (both options configurable as on/off)
- Build a basic React front-end to view and interact with the output (original data passed to LLM and the resulting information)
- The UI will eventually grown in complexity so should be open for extending/scalable
All of this infrastructure (including the LLM) should be included in the Docker configuration.
Should be built with eventual scale in mind and flexibility to run 1000s of documents through at a later stage.
Flexibility to swap out the LLM being used and even store the different results that different installed LLMs have returned.