Rishitha Reddy Mekala

Integrated Gemini-based LLM services for response classification, follow-up dialogue, summaries, and nurse-facing workflows, improving classification latency from about 7.0 seconds to 1.8 seconds. Built validation, retry, structured error handling, and auditable session data workflows that reduced error rates by approximately 50% and maintained about 99.9% crash-free sessions.

Designed and maintained a Node.js/Express/PostgreSQL production platform with 25–40 RESTful APIs supporting approximately 150–200 faculty and staff. Built a Python/Tkinter text-to-speech desktop application with edge-tts, PyInstaller, and ffmpeg for 40+ users, packaging runtime dependencies into a portable Windows executable.

Developed and deployed production changes across 15+ Magento 2 pages using PHP, JavaScript, HTML, and CSS with 100% deployment accuracy. Integrated the Liquifire Orb personalization API across 50+ SKUs, reducing manual CAD customization by approximately 40% and accelerating affected order workflows.

Designed and trained a CNN using Python, TensorFlow, and Keras to predict optical aberration coefficients, achieving 92% held-out prediction accuracy. Generated 3,000+ synthetic labeled images and built NumPy/OpenCV preprocessing and augmentation pipelines.

Built an end-to-end Python ML pipeline across 18,208 satellite and drone image segments, achieving 95% vegetation-classification accuracy with Random Forest. Implemented OpenCV preprocessing, HSV feature extraction, morphology, edge detection, and Hough Line Transform methods for infrastructure analysis.

Designed, developed, tested, and deployed a responsive multi-page website using HTML5, CSS3, and JavaScript from requirements through production delivery. Integrated scheduling and lead-capture functionality and optimized responsive behavior across desktop and mobile devices.

Grade: 3.8

Pursuing a Master of Science in Computer Science with a CGPA of 3.8, expected to graduate in May 2026.