lvvignesh2122 commited on
Commit
40bb6e9
·
1 Parent(s): e4f1c21

Rebrand project to NexusGraph AI

Browse files
Files changed (2) hide show
  1. README.md +17 -8
  2. frontend/index.html +1 -1
README.md CHANGED
@@ -1,5 +1,5 @@
1
  ---
2
- title: Gemini Rag Fastapi Pro
3
  emoji: 🧠
4
  colorFrom: blue
5
  colorTo: indigo
@@ -7,17 +7,18 @@ sdk: docker
7
  pinned: false
8
  ---
9
 
10
- # 🧠 Agentic RAG System
11
 
12
  > **High Distinction Project**: An advanced "Agentic" Retrieval-Augmented Generation system that uses Graph Theory (LangGraph), Structural Retrieval (SQL), and Self-Correction to answer complex queries.
13
 
14
  ## 🚀 The "Master's Level" Difference
15
 
16
  Unlike basic RAG scripts that just "search and dump," this system acts like a **Consulting Firm**:
17
- 1. **Supervisor Agent**: Decides *which* tool to use (PDF, Web, or SQL).
18
- 2. **Self-Correction**: If the answer is bad, the agent *rewrites the query* and tries again.
19
- 3. **Hybrid Retrieval**: Combines **Unstructured Data** (PDFs) with **Structured Data** (SQL Database).
20
- 4. **Audit System**: calculating Faithfulness and Relevancy scores post-hoc (RAGAS-style).
 
21
 
22
  ---
23
 
@@ -45,12 +46,20 @@ The system can now answer quantitative questions like *"Who pays the highest fee
45
  ### 2. 🛡️ Resilience (Circuit Breaker)
46
  If the Google Gemini API quota is exceeded (`429`), the system catches the error and returns a graceful "System Busy" message instead of crashing (`500`).
47
 
48
- ### 3. 🧪 Automated Testing
 
 
 
 
 
 
 
 
49
  Includes a `tests/` suite:
50
  * `test_api.py`: Integrations tests for endpoints.
51
  * `test_rag.py`: Unit tests for retrieval logic.
52
 
53
- ### 4. 🐳 Dockerized
54
  Fully containerized for "Run Anywhere" capability.
55
 
56
  ---
 
1
  ---
2
+ title: NexusGraph AI
3
  emoji: 🧠
4
  colorFrom: blue
5
  colorTo: indigo
 
7
  pinned: false
8
  ---
9
 
10
+ # 🧠 NexusGraph AI
11
 
12
  > **High Distinction Project**: An advanced "Agentic" Retrieval-Augmented Generation system that uses Graph Theory (LangGraph), Structural Retrieval (SQL), and Self-Correction to answer complex queries.
13
 
14
  ## 🚀 The "Master's Level" Difference
15
 
16
  Unlike basic RAG scripts that just "search and dump," this system acts like a **Consulting Firm**:
17
+ 1. **Supervisor Agent (Hybrid)**: Uses **Gemini 2.5 Flash Lite** (Fast) to decide *which* tool to use (PDF, Web, or SQL).
18
+ 2. **Responder Agent (Expert)**: Uses **Gemini 3 Flash Preview** (Smart) to synthesize the final answer.
19
+ 3. **Self-Correction**: If the answer is bad, the agent *rewrites the query* and tries again.
20
+ 4. **Hybrid Retrieval**: Combines **Unstructured Data** (PDFs) with **Structured Data** (SQL Database).
21
+ 5. **Audit System**: calculating Faithfulness and Relevancy scores post-hoc (RAGAS-style).
22
 
23
  ---
24
 
 
46
  ### 2. 🛡️ Resilience (Circuit Breaker)
47
  If the Google Gemini API quota is exceeded (`429`), the system catches the error and returns a graceful "System Busy" message instead of crashing (`500`).
48
 
49
+ ### 3. ⚖️ Hybrid Agent Architecture
50
+ Optimized for Speed and Intelligence:
51
+ * **Routing**: Handled by lightweight `gemini-2.5-flash-lite`.
52
+ * **Reasoning**: Handled by powerful `gemini-3-flash-preview`.
53
+
54
+ ### 4. 🚀 CI/CD Pipeline
55
+ Automated deployment from GitHub to Hugging Face using **GitHub Actions**. Commits to `main` are instantly verified and deployed to production.
56
+
57
+ ### 5. 🧪 Automated Testing
58
  Includes a `tests/` suite:
59
  * `test_api.py`: Integrations tests for endpoints.
60
  * `test_rag.py`: Unit tests for retrieval logic.
61
 
62
+ ### 6. 🐳 Dockerized
63
  Fully containerized for "Run Anywhere" capability.
64
 
65
  ---
frontend/index.html CHANGED
@@ -4,7 +4,7 @@
4
  <head>
5
  <meta charset="UTF-8">
6
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
7
- <title>Gemini RAG Assistant</title>
8
 
9
  <!-- Fonts -->
10
  <link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800&display=swap" rel="stylesheet">
 
4
  <head>
5
  <meta charset="UTF-8">
6
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
7
+ <title>NexusGraph AI - Enterprise RAG</title>
8
 
9
  <!-- Fonts -->
10
  <link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800&display=swap" rel="stylesheet">