Created README file.
Browse files
README.md
CHANGED
@@ -1,14 +1,102 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
---
|
2 |
-
|
3 |
-
|
4 |
-
|
5 |
-
|
6 |
-
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
|
11 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
12 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
13 |
|
14 |
-
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
|
|
|
1 |
+
# π Personal Finance Investment Strategist
|
2 |
+
## Agentic Demo Showcase β Track 3 Entry
|
3 |
+
|
4 |
+
> **Agent-Based Financial Advisor Using LlamaIndex, Nebius AI, and Economic Data**
|
5 |
+
|
6 |
+
Built by: **[Devavrat Samak]**
|
7 |
+
Tech Stack: LlamaIndex | meta-llama/Meta-Llama-3.1-70B-Instruct-fast (Nebius) | Modal Labs | Gradio | World Bank & Yahoo Finance APIs
|
8 |
+
|
9 |
+
---
|
10 |
+
|
11 |
+
## Hackathon Track
|
12 |
+
|
13 |
+
This submission is for:
|
14 |
+
**Track 3: Agentic Demo Showcase**
|
15 |
+
agent-demo-track
|
16 |
+
We use a multi-agent system to:
|
17 |
+
1. Analyze user financial profile
|
18 |
+
2. Enrich with demographic insights
|
19 |
+
3. Fetch live economic indicators
|
20 |
+
4. Generate personalized investment strategy
|
21 |
+
|
22 |
+
---
|
23 |
+
|
24 |
+
## π Overview
|
25 |
+
|
26 |
+
This agentic system provides intelligent, context-aware investment strategies based on:
|
27 |
+
|
28 |
+
| Feature | Description |
|
29 |
+
|--------|-------------|
|
30 |
+
| User Profile | Age group, income, expenses, risk tolerance, goal, timeframe |
|
31 |
+
| Demographic Insights | Life stage, liquidity needs, planning horizon |
|
32 |
+
| Economic Analysis | Inflation, interest rates, GDP growth, market trends |
|
33 |
+
| Tax Awareness | Country-specific tax rules from India and G7 countries, i.e. USA, UK, Canada, Germany, France, Italy, Japan |
|
34 |
+
| Strategy Generation | Asset allocation, recommended instruments, monthly savings plan |
|
35 |
+
|
36 |
+
---
|
37 |
+
|
38 |
+
## How It Works: The Agentic Framework
|
39 |
+
|
40 |
+
### Agent Flow
|
41 |
+
|
42 |
+
Each agent uses natural language reasoning and external tools to build the final strategy.
|
43 |
+
|
44 |
+
### Agents
|
45 |
+
|
46 |
+
| Agent | Role |
|
47 |
+
|-------|------|
|
48 |
+
| **Input Analyzer** | Parses free-form input into structured profile |
|
49 |
+
| **Demographic Profiler** | Maps life stage, liquidity needs, planning horizon |
|
50 |
+
| **Net Worth Checker** | Identifies existing investments and assets |
|
51 |
+
| **Financial Literacy Detector** | Determines complexity level of advice |
|
52 |
+
| **Economic Analyst** | Gathers inflation, interest rate, GDP, and market trend |
|
53 |
+
| **Strategy Advisor** | Generates final investment plan |
|
54 |
+
|
55 |
+
---
|
56 |
+
|
57 |
+
## Technical Architecture
|
58 |
+
|
59 |
+
### Core Components
|
60 |
+
|
61 |
+
| Component | Purpose |
|
62 |
+
|----------|----------|
|
63 |
+
| LlamaIndex | Manages agent orchestration and reasoning flow |
|
64 |
+
| Qwen3-32B on Nebius | Generates human-readable investment strategy |
|
65 |
+
| World Bank API | Provides inflation and GDP data |
|
66 |
+
| Yahoo Finance | Detects market trend (bullish/bearish) |
|
67 |
+
| Hardcoded Interest Rates | For major countries (USA, India, UK, etc.) |
|
68 |
+
| Country Tax DB | Applies local tax logic to optimize portfolio |
|
69 |
+
| Modal Labs | Hosts backend as ASGI endpoint |
|
70 |
+
| Gradio | Frontend interface on Hugging Face Spaces |
|
71 |
+
|
72 |
---
|
73 |
+
|
74 |
+
## Features
|
75 |
+
|
76 |
+
| Feature | Status |
|
77 |
+
|--------|--------|
|
78 |
+
| β
Multi-Agent Workflow | LlamaIndex-based |
|
79 |
+
| β
Country Support | India, USA, UK, Canada, Germany, France, Italy, Japan |
|
80 |
+
| β
Economic Indicators | Inflation, Interest Rate, GDP Growth, Market Trend |
|
81 |
+
| β
Tax-Aware Advice | Yes β country-specific |
|
82 |
+
| β
Risk Tolerance Mapping | Conservative / Moderate / Aggressive |
|
83 |
+
| β
Markdown Strategy Output | Clean, formatted response |
|
84 |
+
---
|
85 |
+
|
86 |
+
## πΉ Video Walkthrough
|
87 |
+
|
88 |
+
π¬ [Insert your video link here explaining how the agents work together]
|
89 |
+
|
90 |
---
|
91 |
+
- Credits
|
92 |
+
- Built with β€οΈ by [Devavrat Samak]
|
93 |
+
- For Hugging Face Gradio Hackathon 2025 β Track 3: Agentic Demo Showcase
|
94 |
+
|
95 |
+
- GitHub Repo: [link-to-your-repo] # Will paste it later on!
|
96 |
+
- Model: meta-llama/Meta-Llama-3.1-70B-Instruct-fast (Nebius AI)
|
97 |
+
- Backend: Modal Labs
|
98 |
+
- Frontend: Hugging Face Gradio
|
99 |
+
|
100 |
+
Thank You!
|
101 |
+
If you're reading this β thank you for checking out our project. We hope you found it insightful and well-built.
|
102 |
|
|