Golang Enthusiast • ML/DL Researcher • Cloud & DevOps Explorer

• Deep Learning & LLMs: Building end-to-end AI pipelines with PyTorch Lightning and Hugging Face Transformers to supercharge real-time recommendation and analytics solutions.
• MLOps & Cloud: Working on AWS & GCP for serverless deployments and large-scale data processing with MLflow, Docker, and Kubernetes.

• AI & ML Frameworks: PyTorch, TensorFlow, Hugging Face Transformers
• Large Language Models: LLaMA, Mistral, GPT-based models, RASA
• Cloud/DevOps: AWS (EC2, Lambda, Fargate, CloudFormation), GCP (Compute Engine, Vertex AI), Azure, Docker, Kubernetes, CI/CD (GitLab/Jenkins)
• Databases: MySQL, MongoDB, Redis, Milvus, Pinecone, Snowflake, Airflow
• GPU Acceleration: CUDA, cuDNN, multi-GPU data parallel training, model optimization (quantization, pruning)

• HelioSync: A Federated Learning Platform
  Developed a comprehensive federated learning platform enabling decentralized training across multiple clients without sharing raw data. Implemented secure aggregation protocols to ensure data privacy and model robustness. The platform facilitates seamless integration with existing machine learning workflows, promoting collaborative model training across organizations.

• Underwater Object Detection Pipeline
  Engineered a machine learning pipeline tailored for detecting objects in the underwater environment. The pipeline includes data preprocessing with contrast enhancement via CLAHE, color correction using a Wasserstein GAN, and object detection utilizing YOLOv.Additionally, integrated AES encryption for secure handling of detection results.

• Backprop-aganda: Neural Network Training Visualizer
  Developed an interactive visualization tool to monitor and analyze the training process of neural networks. Utilized Jupyter Notebook to create dynamic plots showcasing metrics such as loss convergence, weight distributions, and activation patterns, aiding in the debugging and optimization of complex model.

• Diving deeper into Generative AI with advanced model optimization, parallelism, and few-shot learning.
• Expanding frontend horizons with React to become a more complete Full Stack engineer.

• Golang Backend Services
• Full Stack Applications
• LLM & AI Innovations (Mistral, LLaMA, GPT-based systems)

Ask me anything about AI, Machine Learning, Golang, or Cloud—happy to share what I’ve learned!

• Email: smitsaurabh20@gmail.com
• LinkedIn: Saurabh Suman
• GitHub: saurabh98s

“The best way to predict the future is to invent it.” – Alan Kay

Feel free to fork, star, or open an issue if you find something interesting!
Thanks for stopping by and happy coding!