Sarika Mishra

Worked across multiple high-impact client projects, delivering AI solutions that blended ML models, GenAI systems, and robust MLOps practices, directly supporting business transformation and revenue growth. Banking: Designed fraud detection pipeline with ensemble models (Random Forests, Decision Trees), increasing fraud detection accuracy by 30%. Automated ETL workflows for transactional data with data engineers, reducing pipeline latency by 40%. Deployed Power BI dashboards to risk teams, cutting investigation cycles by 20%. Insurance: Built claims triage model (Logistic Regression + Gradient Boosting) integrated with LLM-based summarizer (AWS Comprehend Medical + GPT) for contextual insights. Exposed models through FastAPI microservices on AWS Lambda for real-time claim decisioning. Delivered solution that reduced manual claim review efforts by 35%. Retail: Developed GenAI chatbot using LangChain + vector search (Milvus), enabling personalized, multi-turn product discovery. Boosted customer conversion rates by 15-20% through contextual, memory-driven conversations. Partnered with marketing to align outputs with sales KPIs. Healthcare: Built ML pipeline using Logistic Regression & Random Forests on EHR data; improved readmission risk prediction by 12%. Designed interpretable features for physicians to understand clinical drivers of readmission. Delivered insights via interactive Power BI dashboard for hospital executives. Healthcare Research: Finetuned domain-adapted BART model on biomedical corpora for summarizing research papers. Deployed solution as a Streamlit app, reducing average literature review time by 40%. Helped medical researchers accelerate discovery and improve research throughput.