Integrated CoinGecko API for real-time market data ingestion, enabling sub-minute price refresh intervals, improving
system data reliability, and supporting comparative trend analysis across several cryptocurrencies and exchanges.
Engineered NLP sentiment analysis pipeline parsing social data and classifying polarity, achieving accuracy exceeding
85% overall in trend detection and flagging anomaly events from user-generated financial discussions and commentaries.
Developed ML regression models using LSTM and GRU architectures, delivering Mean Absolute Percentage Error below
3.6%, matching high industry benchmarks and ensuring stable output across Evolving test datasets and timeframes