Kosana Hema Teja Reddy I'm a Computer Science graduate who builds machine learning systems that solve real problems — not just notebooks that stop at 90% accuracy on a clean dataset. Over the past year, I've worked across the full AI/ML stack: training computer vision models for real-world deployment, building NLP pipelines that process messy human language, and integrating LLMs into products people actually use. A few things I've built: 🔹 A YOLOv9-based PPE detection system for construction sites — trained on 3,000+ custom-labeled images, hitting 91% mAP@0.5, with a real-time CCTV inference pipeline that flags safety violations on the spot. 🔹 An AI Finance Assistant that fine-tunes NLP pipelines for expense categorization and connects LLM APIs to a React dashboard for real-time financial insights — including automated anomaly alerts via Twilio. 🔹 A customer feedback analyzer that classifies sentiment with 92% accuracy and uses LLMs to auto-summarize recurring issues at scale. My toolkit: Python, PyTorch, YOLOv9, OpenCV, NLP, LLM APIs, Flask, React.js, MySQL, Git. I care about the parts of ML that most tutorials skip — data labeling quality, inference speed, and actually shipping a model behind an API
python fullstack trainee
Artificial Intelligence Intern
Bachelors, Software