Priyanshi Chaudhary

At Fractal Analytics, I build and deploy multilingual LLM validation platforms that automate manual workflows across global regions. I architected a GPT-4-based system that automated approximately 90% of validation work, improved validation accuracy by approximately 95%, and increased issue-resolution efficiency by approximately 80%.

I own LLM solutions end to end, from prompt research, experimentation, validation logic, and error analysis through architecture, deployment, monitoring, retraining, and production support. My work combines Python, FastAPI, Azure, PostgreSQL, Cosmos DB, LangChain, fallback models, and Google Translate API.

I also built an unsupervised ranking system with EDA, feature engineering, metric validation, and a Streamlit dashboard, alongside clustering-based customer segmentation pipelines using K-Means and DBSCAN.

Earlier, at IIT Delhi's AIA FSM, I developed a YOLOv8 defect-detection model and experimental workflows combining computer vision with collaborative robots for manufacturing quality control. I'm recognized for technical execution, solution quality, and client impact, including a Client Appreciation Award and top-contributor recognition at Fractal Analytics.