Raghav Konduri

Experience

1. Ai Ml Engineer

   Scale AI

   Jan 2025 — Present · US

   Led the design and deployment of a Retrieval-Augmented Generation (RAG) platform that enables enterprise users to ask natural language questions and get context-aware answers from internal documentation, knowledge bases, and policy repositories. Built end-to-end pipelines for document ingestion, text embedding, and semantic retrieval using FAISS and Scale Atlas, supporting millions of documents and real-time query responses. Fine-tuned LLaMA-2, Falcon, and BART models using LoRA, QLoRA, and PEFT on distributed GPU clusters, cutting model training cycles by 60% and improving recall accuracy for enterprise queries. Integrated Scale Nucleus and GenAI APIs to collect human feedback and build continuous improvement loops for prompt evaluation, response scoring, and annotation-based fine-tuning. Optimized inference speed using TorchScript, ONNX Runtime, and mixed precision (FP16/INT8) quantization, enabling efficient low-latency responses under heavy user load. Collaborated with Scale’s internal safety and compliance teams to integrate prompt sanitization, content moderation, and schema validation, ensuring responses meet SOC 2 and GDPR standards. Conducted A/B testing and prompt tuning experiments through internal experimentation frameworks to measure response quality and user satisfaction, driving 25% better factual accuracy.

Education

• Vidya Jyothi Institute Of Technology

  Bachelor of Technology , Artificial Intelligence

• Rowan University

  Master's degree, Data Science