At AgroInsightX, I built computer vision and data pipelines for precision agriculture, supporting a pilot with 690+ farmers that cut crop diagnosis time from 3 days to under 30 seconds. I also reduced geospatial pipeline runtime by 40% and deployed models through FastAPI APIs.
As a freelance data scientist and machine learning engineer, I developed an NLP text-classification pipeline over 568,000+ reviews using spaCy, GloVe embeddings, and LSTM networks, achieving 90.58% test accuracy. I also delivered classification and forecasting models validated with cross-validation.
At RWS Group, I maintained ground-truth validation protocols and evaluation metrics for computer vision models, improving accuracy by 13.3%. My projects include an offline mobile app for detecting coffee pests and diseases and a reinforcement-learning drone navigation simulation.

