Vyoma Shah
Building domain-specific AI agents with LLMs, RAG, and scalable ML pipelines
M.Sc. Data Science @ Uppsala University • Specializing in Generative AI, LLM Agents & Production ML Systems
Featured Projects
From LLM agents to distributed ML pipelines — building AI systems that scale and deliver real impact.
Konfsutra - AI Shortcut Assistant
Built an AI-powered assistant for 500+ Linux shortcuts using RAG pipeline. Engineered vector embeddings and generative AI to answer natural language queries from man pages with high accuracy.
Personal Analyst AI Agent
Designed and deployed AI agent for trading analytics at Traderware. Implemented RAG and tool mapping to reduce token usage by 50% and improve text-to-SQL accuracy. Built HR evaluation tool with live voice-to-text.
Custom Transformer from Scratch
Implemented a fully functional Transformer with multi-head self-attention and positional encoding. Benchmarked on sequence tasks, evaluated impact of hyperparameter tuning on performance.
Scalable Sentiment Analysis
Built distributed pipeline for 14M+ lyrics achieving 92% accuracy and 40% faster processing. Leveraged Hadoop and Spark with PySpark, MapReduce, and HDFS for scalability.
Research Interests
LLM Agents & RAG
Building intelligent agents with retrieval-augmented generation, tool mapping, and efficient token usage.
Scalable ML Systems
Distributed pipelines, big data processing, and production-ready ML infrastructure.
Deep Learning
Transformers, NLP, and custom model architectures from theory to implementation.
Background & Journey
Currently pursuing M.Sc. in Data Science at Uppsala University (2024-2026), specializing in Machine Learning and Statistics. My academic focus includes Deep Learning, Statistical ML, Data Engineering, and Large Language Models.
I recently completed a Generative AI Internship at Traderware, where I built a Personal Analyst AI agent for trading analytics. The system featured sophisticated RAG pipelines, tool mapping to reduce token usage by 50%, and improved text-to-SQL accuracy. I also developed an HR evaluation tool with live voice-to-text capabilities.
My current research interest is developing domain-specific AI agents that integrate retrieval-augmented generation with specialized knowledge bases. I'm exploring how to build agents that understand domain-specific contexts, tools, and reasoning patterns to deliver accurate, efficient solutions.
🎯 Currently seeking PhD opportunities in LLM agents, domain-specific AI systems, and production ML.
Let's Build the Future of AI
Interested in collaboration, research opportunities, or just want to discuss the latest in LLMs and generative AI? I'd love to connect.