Ghanshyam Chodavadiya

( Machine Learning Engineer )

I work on products which comprise neural networks that solve NLP and Computer Vision problems. This includes dataset gathering, neural network training, the science of making it work, and their deployment in production.

feeling_proud

What I Do?

all the data

Data Science & ML/DL

⚡ Structured, Semi-Structured, and Unstructured data 📝 crunch and visualization from Dataset gathering to serving it.

⚡ Experienced in the AI-driven 🤖 product development cycle.

⚡ Create/Manage Data Processing Pipeline 🎢 and Deep Learning Infrastructure for the product.

⚡ Build, Train, Evaluate, and Test the Machine Learning Models in Tensorflow, Keras, Pytorch, and similar libraries & framework.

⚡ Design NLP and Computer-Vision based deep learning models' architecture with critical details.

⚡ Experienced in working with real-time, gigantic data. 🏭

⚡ Transform the latest research (SOTA) in AI into the product.

heatmap

Science behind AI

🎨 Apply mathematics to create neural networks that can work as an inspired system. Yes, I do write code for all matrix manipulation.

🚼 Do babysitting of machine learning models till it becomes production-ready.

🚃 I do write code for training models with multiple GPUs, hyperparameter search, babysitting of training, inference, model profiling, model debugging, optimize for hardware.

cloud hosting

Backend & Cloud

⚡ Create backend for Machine Learning models to serve in a highly scalable environment.

⚡ Manage all databases or additional support to create the backend of the product that also comprises machine learning models.

⚡ Deploy backend on cloud services with multi GPUs in a highly optimized way.

design tools

AI Infrastructure

⚡ Create CI/CD infrastructure development for machine learning engineers to train, test, and fluidly deploy the models.

⚡ An Infrastructure that can monitor model performance, the model bias in inference, and version control of the models.

⚡ Plug and play of models and backend configuration with slack and similar services to control the production setup without touching the code.