AI Engineer
I started out wanting to deeply understand how AI systems actually work, not just use them. That led me to build transformer models from scratch in PyTorch, which gave me a strong foundation in model internals.
From there, I moved into building real-world LLM systems. I developed a Legal Assistant using a RAG pipeline with LangChain, FAISS, and LLaMA-3, where I focused on retrieval quality, prompt design, and system reliability.
At Adya.AI, I worked on designing GenAI pipelines that convert natural language into structured DSL rules. That experience pushed me to think beyond just models and focus on building end-to-end systems—handling parsing, evaluation, and scalability.
What makes me unique is that I combine low-level understanding of models with practical system-building. I don’t just call APIs—I try to understand, optimize, and improve the full pipeline. I also move quickly when learning new tools or domains, which has helped me take ideas from concept to working systems efficiently.