At Gramener, I delivered ML and GenAI work for client projects, including an offer-optimization PoC with patron segmentation, propensity models, and SHAP-based simulation. The simulator projected 16.1% higher theoretical win with 9.6% lower offer spend on validation data.
I also improved GPT-4o mini data-extraction accuracy from 75% to 91%, built RAG pipelines, and spearheaded a LightGBM coupon-targeting model that reached 85% accuracy in the PoC and is now deployed in production. My technical projects include multi-agent RAG and QLoRA fine-tuning.

