⚡ NEXA

AI-Powered Electronics Engineering Agent

Built with

---

🎯 The Problem

Electronics engineering students and hobbyists face a steep learning curve. Debugging a faulty circuit often requires expensive equipment, expert knowledge, or hours of trial-and-error. Traditional LLMs can "help," but they often hallucinate incorrect component values or dangerous configurations.

💡 Our Solution

Nexa is not just another chatbot. It's an agentic AI platform that combines the reasoning power of Google Gemini 3 with a deterministic Physics Validation Engine. It doesn't just generate text—it designs, simulates, validates, and explains complex electronic circuits.

---

🤖 The Agent Swarm

Nexa is powered by a collaborative swarm of 7 specialized AI agents, all coordinated by the Orchestrator.

#	Agent	Location	Unique Capability
1	Orchestrator	All Pages	Intent routing & agent coordination
2	Design Agent	Chat, Schematic	Generates schematics, BOM & firmware from text
3	Diagnostic Agent	Troubleshoot, Analyzer	Root-cause analysis using physics reasoning
4	Simulation Agent	Analyzer, PCB	SPICE-level Bode plots, transient analysis
5	Code Agent	Code Editor	Firmware generation for ESP32/Arduino
6	Vision Agent	PCB, Schematic	Schematic-to-Netlist from images
7	Component Agent	Components	Smart component recommendations

---

✨ Core Capabilities

🔧 1. Agentic Circuit Design

Give Nexa a vague idea, and it generates a complete engineering package.

Input: "Design an RGB LED circuit with ESP32 for color mixing"

Output:

• ✅ ASCII Schematic Diagram
• ✅ Simulation Results (DC Bias, Peak Current, PWM Frequency)
• ✅ Bill of Materials (BOM)
• ✅ Production-ready C++ Firmware

🧠 2. Physics-Validated Debugging

Nexa doesn't guess. It calculates using Ohm's Law, Kirchhoff's Laws, and component datasheets.

Example: User connects 10kΩ resistor to LED.

• Nexa Analysis: I = V/R = 3.3V / 10kΩ = 0.33mA → Dim!
• Nexa Fix: Recommends 220Ω for 15mA optimal brightness.

👁️ 3. Vision-Powered Analysis

Upload a photo of your breadboard or a schematic image, and the Vision Agent extracts components and connections.

📈 4. AI-Powered Circuit Simulation

No SPICE installation required. The Simulation Agent generates:

• Bode plots (frequency response)
• Transient analysis waveforms
• DC operating points
• Power dissipation calculations

🎓 5. Adaptive Learning System

Nexa adjusts its explanations based on your skill level.

• Beginner: "A resistor limits current like a narrow pipe limits water..."
• Advanced: "Apply Thevenin's theorem to simplify the network..."
• Viva Mode: Generates interview-style Q&A to test your knowledge.

---

🧩 10+ Gemini Function Declarations

Nexa uses Gemini's Function Calling to decide which tool to invoke. These are the core functions available:

Function	Description
analyze_circuit	Deep-dive analysis with fault detection
calculate_component_value	Verified physics calculations (Ohm's Law, etc.)
fetch_datasheet	RAG-powered component specs lookup
fetch_lab_rule	Lab safety rules & best practices
validate_solution	Cross-check calculations before presenting
generate_project_plan	Structured project plan with components
generate_learning_summary	Viva questions & concept explanations
fetch_common_mistake	Proactive warnings for known pitfalls
get_user_profile	Retrieve skill level & learning history

---

🏗️ Technical Architecture

graph TD
    subgraph Frontend
        A[React/Vite UI]
    end
    subgraph Orchestrator
        B[Node.js/Express API]
    end
    subgraph AI Engine
        C[Python/FastAPI]
        D[Design Agent]
        E[Simulation Agent]
        F[Vision Agent]
        G[Validation Engine]
    end
    subgraph External
        H[Gemini 3 API]
        I[(MongoDB)]
        J[RAG Knowledge Base]
    end

    A <-->|REST API| B
    B <-->|Internal API| C
    C --> D & E & F
    D & E & F --> G
    C <-->|GenAI SDK| H
    B <--> I
    C --> J

Loading

Tech Stack

Layer	Technologies
Frontend	React 18, TypeScript, Tailwind CSS, Vite
Orchestrator	Node.js, Express, JWT Auth
AI Engine	Python 3.12, FastAPI, Google GenAI SDK
Database	MongoDB
Knowledge Base	YAML-based RAG (Datasheets, Lab Rules)

---

🚀 Quick Start

Prerequisites

• Python 3.9+
• Node.js 18+
• MongoDB (running at localhost:27017)

Installation

# 1. Clone the repository
git clone https://github.com/your-repo/nexa.git
cd nexa

# 2. Configure Environment
# Add your GEMINI_API_KEY to backend/.env

# 3. Run the platform
chmod +x run.sh
./run.sh

Access Points

Service	URL
📱 App	http://localhost:5173
🧪 Test Lab	http://localhost:5173/test-lab
📡 API	http://localhost:5000

---

📂 Project Structure

nexa/
├── backend/
│   ├── api/              # FastAPI Endpoints
│   ├── services/         # AI Agents (Design, Vision, Simulation)
│   ├── functions/        # Gemini Function Declarations
│   ├── knowledge_base/   # RAG Data (Datasheets, Lab Rules)
│   └── validators/       # Physics Validation Engine
├── frontend/
│   ├── src/pages/        # React Pages (Chat, PCB Viewer, Test Lab)
│   └── src/components/   # UI Components
├── docs/                 # Architecture & Planning Docs
├── run.sh                # One-click startup script
└── README.md             # You are here!

---

📚 Documentation

Document	Description
User Manual	End-user guide for all features
Testing Guide	How to use the Test Lab
Architecture Roadmap	System design & future plans

---

⚖️ License

This project is licensed under the MIT License.

---

Engineering Intelligence for Everyone

© 2026 Nexa Platform