R Ajay Kumar

I’m a Computer Science B.E. student and current Machine Learning Intern at Likwid AI, where I architect and deploy an optimized machine learning prediction engine for a core ERP platform. I’ve collaborated on database optimization initiatives to streamline storage and improve data management efficiency, and I’m developing a Retrieval-Augmented Generation (RAG) pipeline to bring advanced NLP and search capabilities directly into the ERP environment.

My projects reflect how I like to work: building end-to-end systems with strong engineering discipline. I designed an autonomous algorithmic trading platform with concurrent services and real-time sentiment using FastAPI and HuggingFace FinBERT, created an Intel XPU-targeted KIVI 2-bit LLM quantization engine using C++/SYCL and oneAPI, and delivered a multilingual RAG chat system deployed on Azure Container Apps. I also enjoy applied LLM work like fine-tuning Gemma-2-2B for Text-to-SQL with QLoRA, and I contribute to open source with a focus on mentoring and community impact.