alolika bhowmik

I’m an AI Research Associate and Data Science postgraduate focused on LLM evaluation, model output quality, and data-driven workflow reliability. I use structured prompts and response frameworks to make model behavior consistent, auditable, and measurable.

In five large-scale AI/LLM evaluation projects, I improved model accuracy, consistency, and instruction alignment by 30%+. I also engineered and validated structured JSON-based data workflows with strict formatting requirements, achieving 95%+ schema accuracy and reducing downstream processing errors.

My work includes high-volume model evaluation, annotation, and quality assurance (1,000+ data points/week). I identify logical gaps, execution errors, and workflow inefficiencies, and I’ve sustained a 95%+ quality benchmark while expanding from model rating to end-to-end interaction auditing and workflow validation.

I’m also a published researcher on ML-based risk prediction and hold two filed patents in applied technology. Alongside research, I’ve served as a Product Owner aligning technical delivery with business goals and built full-stack features and analytics (Power BI dashboards) to turn raw data into actionable product insight.