N.
All work

VENICVoice Enabled Programming Assistant

07 / 07
07
Year
2023
Role
Core Developer & System Architect
Duration
12 months
Client
Independent

Live project Source

Overview

Developed an AI-powered voice-controlled programming assistant enabling hands-free code development. Built a dialogue management system on RASA 3.0.8 with 200+ intent classifications and real-time audio feedback under 500ms.

Challenge

Traditional coding requires extensive keyboard use, creating challenges for developers with RSI or physical limitations, while repetitive coding tasks consume valuable development time.

Solution

Built a dialogue management system using RASA NLU with 200+ programming intents, Azure Cognitive Services for speech recognition, a real-time audio feedback module, and published a production-ready Python package to PyPI.

Results

  • 01

    95%+ intent classification accuracy

  • 02

    <500ms audio feedback latency through TTS pre-generation

  • 03

    40% reduction in code-writing time for repetitive tasks

  • 04

    1000+ package installations from PyPI

  • 05

    Deployed to Okteto Cloud and Azure Kubernetes

Technology

PythonRASA 3.0.8ReactAzure Cognitive ServicesTensorFlowFlaskDockerKubernetesWebSocketGitHub Actions