Sai Keerthana Terala
Professional Summary
B.Tech Information Technology student (expected July 2026) with hands-on experience in deep learning, computer vision and Python development. Completed internships in Data Science and Software Development where I built ML prototypes (classification, anomaly detection, recommendation/system analysis) and production-oriented Python applications (web scraper, algorithm implementations). Strong practical experience with TensorFlow/Keras, YOLOv8, EfficientNet, OpenCV, and model interpretability (GradCAM). Seeking internship or entry-level roles in Software Engineering, Python Development, Data Analytics, or AI/ML where I can apply model development, data pipelines, and CV expertise to deliver measurable results.
Technical Skills
Work Experience
- Designed and implemented prototype anomaly detection pipelines using Python, Pandas and scikit-learn to identify outliers in structured datasets; documented evaluation metrics and recommended feature improvements for productionization.
- Developed sentiment analysis classifiers on text datasets using TensorFlow/Keras and scikit-learn preprocessing; measured performance via confusion matrix, precision, recall and F1 to guide model selection.
- Built a proof-of-concept recommendation system combining content-based features and collaborative signals using Python and Pandas to generate ranked item suggestions for small-scale datasets.
- Performed model validation and comparative evaluation with cross-validation and holdout sets; produced clear model comparison reports to assist technical stakeholders in choosing deployment candidates.
- Collaborated with mentors in guided sessions to convert ML prototypes into deployment-ready designs, documenting input/output contracts, performance targets, and monitoring considerations.
- Managed code and experiment artifacts using Git and GitHub, maintained reproducible notebooks, and presented findings in sprint demos following Agile practices.
- Developed four Python applications (Temperature Converter, Number Guessing Game, Sudoku Solver using backtracking, and a Web Scraper) demonstrating end-to-end design, implementation and testing.
- Implemented a Sudoku solver using an optimized backtracking algorithm in Python, reducing search time with heuristic ordering and validating correctness across multiple puzzle sets.
- Built a resilient web scraper to extract structured product data from 100+ e-commerce pages, implementing pagination handling, retry logic and error recovery to maximize data completeness.
- Packaged scraper outputs into clean CSV/JSON datasets suitable for downstream analysis and model training; documented data schema and preprocessing steps for reuse.
- Managed the project codebase with Git, participated in Agile sprint cycles, and used PR-based reviews to maintain code quality; received a Letter of Recommendation upon internship completion.
- Prepared developer documentation and delivered sprint demos to mentors, translating user stories into deliverable tasks and iterating on feedback to improve application robustness.
Projects
- Built a deep learning classifier to categorize brain MRI scans into four classes (Glioma, Meningioma, Pituitary, No Tumor) using EfficientNet-B0 and transfer learning on a dataset of 7,023 images.
- Achieved 97% test accuracy and evaluated model performance with confusion matrix, precision, recall and F1-score; integrated GradCAM to produce interpretable heatmaps highlighting model attention over tumor regions.
- Developed a real-time CCTV surveillance pipeline to detect five vehicle classes using pretrained YOLOv8 (COCO weights) and implemented centroid-based multi-object tracking with unique ID assignment.
- Implemented virtual line-crossing logic and timestamp-delta based speed estimation achieving 90.91% estimation accuracy with a 3.27 km/h mean absolute error, providing a sensor-free alternative to radar systems.
Education
Certifications
Achievements
- Adobe India Hackathon — Qualified for Round 2 — 2024
- IndustrAI 24-Hour Hackathon — Participant: IIT Madras
Powered by Drivetube · Create your own profile at drivetube.ai