Understanding the Code

This document explains the key components and functionality of the code that integrates with the Ollama LLM (Large Language Model) for analyzing resumes and job descriptions.

Overview

The code is structured to:

1. Load environment variables.
2. Initialize the Ollama LLM with a specified model.
3. Provide utility functions for interacting with the LLM.
4. Analyze job descriptions and resumes to produce insights.

Key Components

1. Loading Environment Variables

from dotenv import load_dotenv
load_dotenv()

Purpose: Loads environment variables from a .env file into the application. These variables include the model name for the LLM.

Example:

.env file:
OLLAMA_MODEL=llama3.2:latest

2. Initializing the LLM

LLM_MODEL = os.environ.get("OLLAMA_MODEL", default="llama3.2:latest")
llm = OllamaLLM(
    model=LLM_MODEL,
)

• LLM_MODEL: Retrieves the model name from the environment variables or uses "llama3.2:latest" as the default.- OllamaLLM: Initializes the LLM using the specified model.

3. Templates for Prompts

from .templates import OLLAMA_TEMPLATES

• OLLAMA_TEMPLATES: Contains predefined templates for different tasks such as job description validation and resume analysis.Templates define the structure of the input prompts provided to the LLM.

4. Invoking the LLM

def invoke_llm(template_name: str, inputs: dict) -> str:

This helper function:

• Takes a template name and input variables.
• Validates the template name.
• Constructs a PromptTemplate using the template and inputs.
• Executes the chain combining the prompt and the LLM.
• Returns the response from the LLM.

Example Workflow

Input:

invoke_llm("RESUME_ANALYSIS", {"resume": "Sample Resume", "job_description": "Sample Job Description"})

Output: The LLM’s analysis of the resume against the job description.

5. Job Description Validation

def analyze_job_description(job_description: str) -> bool:

Purpose: Checks if the job description is valid.

Process:

1. Ensures the input is non-empty.
2. Invokes the LLM with the JOB_DESCRIPTION template.
3. Validates that the LLM’s response is either True or False.

Returns: True if the job description is valid, False otherwise.

Example:

• Input:

  analyze_job_description("We are looking for a Python developer with Django experience.")
  

• Output:

  True
  

6. Resume Analysis

def analyze_resume(full_resume: str, job_description: str) -> str:

Purpose: Analyzes a resume against a job description.

Process:

1. Ensures both inputs are non-empty.
2. Invokes the LLM with the RESUME_ANALYSIS template.

Returns: A detailed analysis string from the LLM.

Example:

• Input:

  analyze_resume(
      "Experienced Python developer with Django expertise.",
      "Looking for a Python developer with Django and Flask experience."
  )
  

• Output:

  A string containing the LLM’s analysis of the resume compared to the job description.
  

Error Handling

Invalid Template Name:

if template_name not in OLLAMA_TEMPLATES:
    raise ValueError(f"Invalid template name: {template_name}")

Ensures only valid templates are used.

Empty Input Validation:

• analyze_job_description: Ensures the job description is non-empty.
• analyze_resume: Ensures both resume and job description inputs are non-empty.

Unexpected Responses:

if response not in ["True", "False"]:
    raise RuntimeError(f"Unexpected response from LLM: {response}")

Ensures the LLM’s response is predictable.

Summary

This code provides a robust framework for leveraging the Ollama LLM to:

1. Validate job descriptions.
2. Analyze resumes against job descriptions.
3. Return structured and actionable insights.

The use of templates ensures consistency in prompts, while helper functions like invoke_llm simplify interaction with the LLM. Error handling and input validation ensure reliability and robustness.