Abhinav Sharma

As a data scientist and AI engineer with experience spanning fintech, retail, healthcare, and insurance, I specialize in building end-to-end machine learning, Generative AI, and data-driven systems that bridge the gap between business strategy and technical execution. At Capgemini’s Advanced Analytics & AI team, I designed, deployed, and scaled predictive and generative AI solutions leveraging Azure cloud, MLflow, and Python ecosystems. My work focused on developing intelligent, production-grade ML systems and LLM-driven applications that delivered measurable business impact.

I have led multiple full-cycle ML projects — from data preprocessing, feature engineering, and statistical modeling to MLOps deployment. I built models for fraud detection, churn prediction, and demand forecasting using algorithms such as Logistic Regression, Random Forests, XGBoost, and LSTMs. I implemented experiment tracking and CI/CD workflows using MLflow and Azure ML, ensuring reproducibility, version control, and continuous model improvement. Through techniques like feature selection, Bayesian optimization, and uncertainty quantification, I enhanced model accuracy and generalization.

In the Generative AI space, I developed and deployed Azure OpenAI-based solutions, including LLM-powered chatbots, text summarizers, and retrieval-augmented generation (RAG) systems. I engineered hybrid search and conversational AI systems that used vector embeddings, memory components, and adaptive prompts to provide contextual, enterprise-ready performance. I also built an agentic Text-to-SQL system integrating LangChain, Python, and Azure APIs to convert natural queries into optimized SQL — achieving over 90% accuracy and cutting manual workload by 70%.

Beyond modeling, I have delivered large-scale data pipelines and semantic search systems that combine structured and unstructured data, using MySQL, Power BI, and Azure data services for analytics and visualization. As a technical lead, I guided cross-functional teams, mentored junior data scientists, and aligned project delivery with client and business objectives.

Earlier, at Snapmint Credit Advisory, I developed ML-driven credit scoring models using alternative and behavioral data to assess creditworthiness and reduce default rates. I also built ad performance and ROI models using uplift modeling and marketing mix analysis, improving campaign ROI by 40% and reducing acquisition costs by 25%. My contributions in credit risk modeling, real-time scoring APIs, and fraud detection strengthened Snapmint’s core decisioning systems.