Sameer Jathavedan

Background

About Sameer Jathavedan

I specialize in artificial intelligence and software engineering, with a strong focus on deep learning and system architecture. My Master's in AI, combined with hands-on experience at Zenseact AB and Ericsson, has allowed me to bridge the gap between cutting-edge AI research and real-world applications. At Zenseact, my thesis work involves developing a patent-pending deep learning solution that uses Graph Neural Networks (GNNs) and Transformers to enhance SD Map Localization. By integrating GNSS and INS sensor data, I achieved 97% localization accuracy, significantly improving vehicle navigation in challenging environments. At Ericsson, I worked on Retrieval-Augmented Generation (RAG) solutions and AI-powered agents using LangChain and Amazon Bedrock. Notably, I built an SQL agent capable of translating natural language queries into SQL, enabling non-technical users to interact with databases efficiently. I also developed an XGBoost-based system that classifies fault categories from millions of monthly hardware tests, achieving 92.8% accuracy. Beyond AI, my software engineering experience at OneConvergence Devices Pvt. Ltd. involved developing a Kubernetes-based MLOps platform, optimizing training data management and serverless inference for scalable AI systems. This fusion of deep learning expertise and software engineering skills enables me to design AI solutions that are both innovative and practical. Let’s connect to explore how we can drive the future of AI together.