Lohitasrith Gondi

I am a data scientist with four years of hands-on experience building end-to-end ML and AI systems across GenAI, NLP, recommendation, and computer vision domains. I focus on turning cutting-edge research into production-grade solutions that deliver measurable business impact.

At Capital One, I built LangChain-based AI agents, GraphRAG pipelines, and scalable SageMaker deployments, reducing resolution time by 35% and improving answer accuracy through A/B tests and human-in-the-loop feedback. I also implemented model explainability and bias audits aligned with ResponsibleAI practices.

Previously at Dixon Technologies I developed predictive maintenance, customer segmentation, and defect-detection systems, deployed quantized CV models to NVIDIA Jetson, and implemented serverless architectures that reduced costs and improved scalability. In academic settings I fine-tuned LLMs with QLoRA/PEFT and led RAG and evaluation workflows that raised relevance and contextuality metrics substantially.

I thrive on designing low-latency inference pipelines, reproducible MLOps workflows, and user-centric GenAI solutions using tools like FastAPI, Docker, MLflow, Spark, LangChain, and AWS. I am passionate about building robust, explainable models that are production-ready and compliant for enterprise use.