Sreenitya Thatikunta is a Computer Science undergraduate at IIT Jodhpur who builds at the intersection of AI, machine learning and software engineering — diffusion and vision-language models, forecasting pipelines, and full-stack web products.
Hello, I'm Sreenitya!
I'm a Computer Science undergraduate at IIT Jodhpur who loves building things at the intersection of AI, machine learning, and software engineering. I enjoy taking ideas from “this would be cool” to something people can actually use, and I'm always excited to explore new technologies if they help solve the problem better.
I like creating products that are simple, fast, and thoughtfully designed, with as much attention to the user experience as the code behind it. Whether it's an AI experiment, a web application, or a side project that started out of curiosity, I enjoy building things that are both technically interesting and genuinely fun to use.
Outside of tech, I'm probably watching Formula 1 or cricket, hunting for a good coffee spot, or pretending I'll watch just one episode before inevitably finishing half the season.
Fine-tuned PixArt-diffusion with LoRA on 3,000+ hand images and LLaVA captions. Reduced FID by 13.5% via Optuna tuning, benchmarked variants with FID and T2I-CompBench, tracked experiments in W&B, and deployed a Gradio demo.
Built a CLIP-guided cross-modal segmentation model for text-conditioned facial occlusion localization. Used synthetic training data, prompt augmentation, and Dice loss optimization to reach 94.4% mIoU on CelebA and 94.1% on LFW.
Built a ResNet-50 image retrieval pipeline for CIFAR-10 with LDA reduction and classifier evaluation. Delivered 91%+ accuracy and deployed the system in Streamlit.
Email sreenityathatikunta@gmail.com, or use the contact form on the homepage.
This site publishes a read-only JSON API, an OpenAPI 3.0.3 specification, markdown variants of every page and an llms.txt index, so agents do not have to scrape the interface.