Chaitanya Terli

I've built AI-powered products including ProjectDNA, a semantic project discovery platform that uses LLaMA, SentenceTransformers, MongoDB, and knowledge graphs to surface related projects and generate enhanced ideas.

As a Machine Learning Research Intern at NIT Andhra Pradesh, I designed and evaluated action anticipation models using EPIC-KITCHENS-100 and EGTEA Gaze+ datasets. I built end-to-end video ML pipelines for preprocessing, feature extraction, training, and inference, evaluating RNN, LSTM, and Transformer architectures with Top-1/Top-5 accuracy, per-class recall, and t-SNE visualizations.

I've also developed a no-code data-cleaning web application and a real-time traffic anomaly detection system using YOLOv5 and Swin Transformer. My work spans Python, FastAPI, PyTorch, TensorFlow, Streamlit, MongoDB, and modern LLM frameworks.

Beyond building, I've published research on automated AI peer review, led a three-day workshop on portfolio development, deployment, and AI tools, and mentored students through ATAL Lab.