At MRX Digital Tech, I productionized machine learning pipelines using Python and SQL, covering data preparation, feature engineering, and statistical modeling. I developed and optimized AdaBoost ensembles and PyTorch deep learning models for predictive analytics and NLP workloads.
I also optimized predictive models through hyperparameter tuning and evaluation, and integrated explainability, real-time inference, automated retraining, and scalable data processing for production applications.
On AgentFlow AI, I built a Planner–Executor–Reviewer workflow and a dual-store RAG pipeline using FAISS, Pinecone, and Sentence Transformers. The reviewer checks for unsupported outputs and triggers self-correction.
I also developed an LLM-powered travel planner and a Jira story automation chatbot. The chatbot automates story generation and backlog documentation, reducing manual effort by 60% and saving 10–15 hours per sprint.
