I built JOBCode, a Django-based job platform with user authentication, job management, application tracking, and resume and profile management. Its recommendation engine matches skills, experience, and resumes to jobs.
I integrated the Adzuna API and Google Gemini, with Celery and Redis supporting background processing and automated job ingestion. I also implemented job deduplication, regression testing, production logging, and cloud deployment.
At 1stop.ai, I developed an end-to-end question-answering model using the Facebook bAbI dataset. I worked on preprocessing, model training, and evaluation, using Keras to learn relationships between questions, supporting context, and expected answers.
For a customer churn prediction project, I applied Random Forest, XGBoost, and neural networks to a dataset of 100,000+ customer records. The deep learning model achieved 80% accuracy, and A/B testing validated strategies that reduced churn by 12%.
