Initialisation of codebase

.gitignore

@@ -0,0 +1,6 @@
+data/raw/*
+.cursor/*
+.vscode/*
+.venv/*
+.git/*
+

README.md

@@ -6,7 +6,7 @@ colorTo: indigo
 sdk: gradio
 sdk_version: 6.17.3
 python_version: '3.12'
-app_file: app.py
+app_file: main.py
 pinned: false
 license: apache-2.0
 short_description: Multilingual AI assistant to guarantee fair pay

configs/models.yaml

@@ -0,0 +1,19 @@
+# Model registry — pin every revision by SHA or explicit tag. Never `latest`.
+# All entries are TBD placeholders; fill in `repo` and `revision` as models
+# are chosen. Read by src/backend/config.py:load_models_config().
+
+llm:
+  default:        # Core LLM (default)
+    repo:     # TBD — e.g. "org/model-name"
+    revision:     # TBD — pin by SHA or tag
+  swap_in:        # Core LLM (swap-in), one-line swap via get_llm_client()
+    repo:     # TBD
+    revision:     # TBD
+
+embeddings:       # directive RAG embeddings
+  repo:     # TBD
+  revision:     # TBD
+
+parsing:          # document parsing (OCR / table extraction)
+  repo:     # TBD
+  revision:     # TBD

main.py

@@ -1,15 +1,10 @@
-import gradio as gr
-import spaces
-import torch
+"""Thin entry point for the Pay Equity Gradio Space.
 
-zero = torch.Tensor([0]).cuda()
-print(zero.device)  # <-- 'cpu' 🤔
+All orchestration lives in src/backend; this file only builds and launches the
+server. HF Space `app_file` points here.
+"""
 
-@spaces.GPU
-def greet(n):
-    print(zero.device)  # <-- 'cuda:0' 🤗
-    return f"Hello {zero + n} Tensor"
+from src.backend.server import build_server
 
-
-demo = gr.Interface(fn=greet, inputs=gr.Number(), outputs=gr.Text())
-demo.launch()
+if __name__ == "__main__":
+    build_server().launch(show_error=True)

pyproject.toml

@@ -7,5 +7,8 @@ requires-python = ">=3.12"
 dependencies = [
     "gradio>=6.17.3",
     "spaces>=0.50.4",
-    "torch>=2.12.0",
+    "torch",
 ]
+
+[tool.uv]
+extra-index-url = ["https://download.pytorch.org/whl/cu118"]