Innovative ML & Generative AI Engineer with 1.5+ years in NLP, Computer Vision, and Deep Learning.
My life’s goal is to develop vision algorithms that function like the human brain—my first step toward human-like robots. I believe that if humans can be created, we can engineer beings with similar perception and cognition, harnessing technology’s full potential.
Currently, designing and building a functional AI office assistant powered by a large language model to automate customer scheduling, estimate and follow-up workflow using n8n.
Developed end-to-end automation and AI pipelines including Python web scrapers, SKU attribute extraction with Gemini API and OCR, AWS deployment (EC2, S3, SQS), RAG-based document processing with LLaMA and ChromaDB, legacy system integration APIs, and research documentation.
Fine-tuned Faster R-CNN & Mask R-CNN for environmental datasets and designed a GAN Brightener + ViT-FastRCNN for low-light imagery. Drafting manuscript on image denoising: rebuilt leading models from scratch to extract design insights and now applying them to create a novel GNN-powered hybrid denoising model
Implemented hallucination detection/mitigation with Google API, T5, and FAVA; built a real-time translator with Whisper & ElevenLabs.
Whisper ASR, GPT-3.5 translation & ElevenLabs TTS pipeline.
ViewBuilt a Retrieval-Augmented Visual Question Answering (RAG-VQA) system that combines image understanding (CLIP) with text retrieval (MiniLM + WIT dataset) using FAISS for large-scale search.
GAN Brightener + ViT-FastRCNN dual-architecture for low-light feature detection.
Mask R-CNN & Faster R-CNN on MS COCO. View
Linear & ensemble regression models achieving R²≈1 for purchase amount forecasting. View
Real-time face detection, recognition & voice feedback with OpenCV, dlib, and Flask.
ViewYOLO, U-Net, and RL agents in CARLA for autonomous navigation.
GPA: 4.0
GPA: 3.26
Email: prathyusha.hyra@gmail.com
Phone: +1 618-818-0226
LinkedIn Google Scholar GitHub