Hands-on Machine Learning Internships

Worked on a machine learning system for Google Search performance and webpage prioritization.

Performed data analysis, feature engineering, data validation, and target-leakage detection on large-scale search data.

Compared ML models including Random Forest, XGBoost, LightGBM, CatBoost, and Logistic Regression.

Used 5-fold client-grouped validation to improve evaluation reliability.

Achieved Precision@50 of 0.444 compared with a 0.392 baseline (+5.2 percentage points).

Built a human-in-the-loop content prioritization workflow for SEO review.

Developed a hybrid AI-based invoice parsing system to extract structured data from PDF, image, Excel, and CSV invoices.

Implemented OCR, NLP preprocessing, regex/rule-based extraction, validation, and LLM-based fallback to improve extraction reliability.

Extracted key invoice fields including supplier/buyer details, invoice information, line items, pricing, GST/tax, and totals, and converted them into structured JSON.

Built FastAPI REST APIs with Pydantic validation and deployed the backend using Docker on Render.

Integrated validation and quality checks to identify missing or inconsistent critical invoice information before generating the final output.

Worked on a machine learning project for heart disease risk prediction using structured/tabular health-related data.

Performed basic data preprocessing, exploratory data analysis, feature preparation, and model training.

Experimented with classification models and evaluated their predictions using standard classification metrics.

Worked with Python and common machine learning/data analysis libraries.

Gained practical experience in the end-to-end machine learning workflow, from data preprocessing to model evaluation.