👋 I'm Vanshika Arora, an AI/ML Engineer specializing in deep learning, model fine-tuning, and productionizing ML systems. 🤖 I work with PyTorch, TensorFlow, custom architectures (CNNs, transformers, attention mechanisms), and advanced training techniques like focal loss and transfer learning. ⚙️ Skilled in model optimization, hyperparameter tuning, deployment pipelines with FastAPI, Docker, AWS, and building scalable ML infrastructure. 🚀 Passionate about taking models from research to production — fine-tuning for real-world performance, reducing latency, and shipping ML systems that scale.
-
18:22
(UTC -12:00) - in/vanshika-arorax
- vanshikaarorax
Pinned Loading
-
NutriGen
NutriGen PublicNutriGen 🧠🥗 An AI-powered RAG Nutrition Chatbot. Built using Next.js, OpenAI, Supabase (pgvector), and Gemma-7B, NutriGen combines local + cloud models for context-aware question answering on nutri…
Jupyter Notebook
-
Verdict-gpt
Verdict-gpt PublicComplete GPT-2 architecture from the ground up including tokenization, embeddings, transformer layers, attention, and training trained on The Verdict novel dataset,120 million parameters,768 dimens…
Jupyter Notebook
-
ai_agent
ai_agent PublicThe AI Agent Recommendation Application is a platform that helps users find the most suitable AI agent based on their specific requirements, such as travel planning, business automation, or custome…
JavaScript
-
Movie-bub
Movie-bub PublicMovie-bub is an app where you can know about all the movies in the world,you can see their trailers,description and many more functionalities are there..
JavaScript
-
CortexAI
CortexAI PublicA robust image classification system using a Convolutional Neural Network (CNN) that would classify if subject has a tumor or not base on MRI scan.
Jupyter Notebook
-
Trading-Crypto-stocks-reccomendation
Trading-Crypto-stocks-reccomendation PublicCrypto Recommendation: Built AI platform to recommend crypto ,added buy/shell recommendation using pinescript and recommendation is cutomizable according to user.
HTML
If the problem persists, check the GitHub status page or contact support.
