prakash P

-Designed, fine-tuned, and deployed enterprise-grade Generative AI solutions to support banking operations, digital channels, and internal decision systems, ensuring security, compliance, and scalability across platforms. -Led LLM fine-tuning and optimization initiatives (prompt tuning, task-specific adapters, domain alignment) for internal financial use cases, improving model efficiency and response accuracy by ~35% across risk, operations, and customer-support workflows. -Implemented Retrieval-Augmented Generation (RAG) systems to enable contextual retrieval from policies, procedures, knowledge bases, and regulatory documents, reducing manual research effort by ~40% and accelerating issue resolution. -Engineered multi-agent AI workflows using LangGraph and AutoGen to automate complex, multi-step banking processes such as document analysis, exception handling, and decision routing, increasing operational productivity by ~30%. -Developed conversational AI solutions (chat and virtual assistants) using OpenAI APIs and LangChain, enabling real-time, accurate responses for customer-facing and internal support use cases, contributing to a ~25% improvement in customer satisfaction metrics. -Integrated vector databases and embedding pipelines to enhance semantic search, document similarity, and RAG context relevance, improving retrieval precision and recall by ~20% across enterprise knowledge systems. -Designed and deployed secure, scalable FastAPI-based microservices for LLM inference and orchestration, enabling low-latency model serving, monitoring, and observability in cloud and hybrid environments. -Collaborated closely with risk, compliance, cybersecurity, and data engineering teams to ensure AI solutions met financial-industry governance, data privacy, and regulatory standards. -Supported enterprise adoption of GenAI by creating reusable frameworks, APIs, and best practices, accelerating AI solution rollout across multiple business units.

-Designed and deployed enterprise ML models for transaction monitoring and risk analysis, improving fraud detection accuracy by 25% and reducing false positives by 15%. -Led migration of legacy risk data systems into AWS and Azure cloud pipelines, improving scalability and performance while lowering costs by 20%. -Integrated NLP-based LLM models into customer personalization workflows, increasing engagement and retention by 18%. -Established MLOps CI/CD pipelines for AI models, reducing deployment cycles from weeks to days. -Provided training and mentorship on ML best practices, driving adoption of Generative AI and LLM solutions across multiple business units.

• Developed and deployed customer feedback classification models using historical data, improving campaign targeting and churn mitigation strategies. • Created a churn prediction model using logistic regression and XGBoost, proactively identifying at-risk customers and enabling targeted retention efforts. • Engineered NLP-driven customer support analysis pipeline to extract sentiment, escalation triggers, and support quality metrics from chat and call transcripts. • Integrated ML scoring services into Salesforce CRM, enabling real-time decisioning and automation of upsell offers via RESTful APIs. • Designed and deployed automated data pipelines using PySpark and Apache Airflow to preprocess and monitor contract, billing, and product usage data. • Enabled marketing analysts to track KPIs in near-real-time by developing dashboards using Power BI and GCP Looker Studio. • Mentored junior engineers on model evaluation strategies, CI/CD best practices, and experimentation workflows. • Deployed models using Vertex AI, ensuring scalable training and inference performance for production workloads.