AI voice Assistant (5 Months)
Efficient Voice Processing : Leverages GTTS for natural text-to-speech and NLP for accurate command interpretation, enabling seamless user interactions.
Task Automation : Handles queries, tasks, and responses via voice commands, reducing manual input by 70% in daily use cases.
Real-Time Responsiveness : Processes inputs with low latency using Python libraries, achieving 85%+ accuracy on diverse accents.
Customizable & Scalable: Built over 4 months with modular design for adding features like reminders or multilingual support
- Used Mlflow for experiment tracking and ASR model monitoring, ensuring continuous improvement through real-time feedback loops and automatic model evaluation.
- Enhanced accessibility by adding emotion-aware response generation using sentiment analysis (RoBERTa), improving user satisfaction metrics (USM) by 25% during pilot testing.
Flight Ticket price predection ( Six Months)
Performed extensive data preprocessing, feature engineering, and outlier handling, incorporating temporal features (seasonality, holidays, weekdays), route-specific factors, and demand-supply indicators to boost model performance by 28%
Accurate Price Forecasting: We have Trained ML models on flight data (distance, stops, duration) to predict prices with 85-90% accuracy via regression techniques.
Feature-Driven Insights: We Analyzed key patterns like journey time and stops using EDA, enabling reliable predictions on real-world datasets.
Scalable Model Performance: We Applied Random Forest/XG Boost for robust handling of variables, reducing prediction errors by 20% on test sets.
Practical Travel Optimization :We Delivered tool for users to anticipate costs, cutting booking expenses through data-driven patterns over 6 months.
- Delivered a 22% improvement in profit margins for the simulated business scenario by enabling early-book pricing recommendations and surge-alert forecasting.