An intelligent coding companion that generates code from natural language prompts and then explains the generated code to the user in an easy-to-understand language. Additionally, it can also assist in debugging and code correction.
In development - Module 1 in progress
Module 1: Natural Language Understanding (NLU) Pipeline
The NLU module is the entry point of CodeBud. Before any code can be generated, the system must first understand what the user is asking — this is Module 1's job.
Given a prompt like:
"Write a Python function to find factorial"
The module extracts two things:
- Intent — what the user wants to do (e.g. generate_code, debug_code, explain_code)
- Slots — the parameters of that intent (e.g. language: Python, task: factorial function)
This structured output is then passed downstream to the code generation engine.
Powered by BiLSTM-CRF trained on the SNIPS NLU dataset.
Module 1 achieves 97.64% intent classification accuracy on the SNIPS test set (1,400 samples, 7 intent classes).
Module 2: Code Generation Engine
Generates syntactically correct, multi-language code from structured intent and slot inputs extracted by the NLU module
Module 3: Code Explanation
Explains generated code line by line in beginner-friendly language
Module 4: Debugging Assistant
Works alongside user to point out errors being generated or in case a source code is given, fixes the errors and return a clean code.
Module 5: Code Optimization
Recommends efficient alternatives to improve execution speed and reduce memory usage
Module 6: Documentation Generator
Automatically produces inline comments, function descriptions, and README-ready API documentation
Module 7: Learning Assistant
Explains programming concepts, provides examples, and recommends resources based on user queries
Module 8: Code Review System
Analyzes code quality, flags code smells, checks naming conventions, and suggests refactoring opportunities