Prakhar Kumar

With over 14 years of experience as a Technical Manager, expertise lies in AI/ML, Generative AI, NLP, and LLM-based solutions driving innovation and automation. Leading high-performing teams, specializes in designing and deploying LLM-powered applications like RAG-based chatbots, intelligent search, and document automation. Leveraging skills in Python and ML Ops, ensures scalable and efficient AI pipelines, while deep learning and NLP abilities enable sentiment analysis, keyword extraction, and development of multilingual AI systems. Proficient in GenAI frameworks like LangChain, AutoGPT, and CrewAI, as well as vector databases such as FAISS and Pinecone, architects end-to-end AI-driven automation. Passion for cutting-edge AI advancements drives continuous integration of LLMs, AI agents, and enterprise AI solutions to enhance decision-making and drive business transformation.

Contract Loading Automation using RAG and GPT-3.5
Optum Global Solution  | 01/2025– 04/2025

• Led the design and development of an intelligent contract loading system to extract structured data from unstructured Word documents.
• Architected a Retrieval-Augmented Generation (RAG) pipeline using OpenAI GPT-3.5 for context-aware information extraction.
• Integrated the AI model output with UiPath bots to enable seamless downstream processing and automation.
• Deployed services using FastAPI and Python, ensuring scalability, performance, and compliance with internal standards.
• Acted as Technical Architect, responsible for solution design, code reviews, and compliance checks, while contributing 40–50% hands-on coding throughout the project lifecycle.

Framework Mapper: Policy Mapping Automation
Optum Global Solutions  | 09/2024– 11/2024

• Designed and architected an AI-powered solution to compare and map internal policy documents with industry standards, focusing on control-level similarity.
• Implemented BERT embeddings and spaCy NLP techniques to extract and measure textual similarity between policy statements.
• Integrated a deep learning-based self-learning model to improve mapping accuracy over time and support continuous learning from user feedback.
• Developed the backend using Python for AI/ML components and .NET Core for the user interface to present results interactively.
• Served as the Technical Architect, leading solution design, model selection, integration, and compliance, while contributing to code development and technical reviews.

TA Quality Process Automation
Optum Global Solution | May 2024 – Jul 2024

• Designed and led the development of an end-to-end automated Quality Assurance process for requisitions received from the Talent Acquisition (TA) team.
• Leveraged Generative AI (GPT) with a RAG-based model to intelligently extract and validate key fields from unstructured free-text inputs.
• Integrated the AI output with a Power Automate bot to perform downstream validation and processing tasks, ensuring accuracy and consistency across requisitions.
• Enabled full automation of the QA workflow, significantly reducing manual effort and turnaround time.
• Served as Technical Lead, responsible for solution architecture, technical oversight, and guiding the development team throughout the project lifecycle.

PVO Process Reengineering & Automation (Provider)
Optum Global Solutions | Nov 2023 – May 2024

• Led the reengineering and automation of the Provider Verification and Update (PVO) process to streamline manual workflows and reduce dependency on outbound calls.
• Automated updates to NDB and OnBase systems using UiPath bots, significantly improving data accuracy and reducing turnaround time.
• Introduced a Provider Verification Link and automated email triggers as alternative data sources, minimizing the need for agent intervention.
• Applied Fuzzy Logic-based ML models to match and normalize provider addresses, ensuring clean and consistent data entry into NDB.
• Served as Tech Lead/Associate Manager, responsible for solution architecture, automation strategy, and team coordination across implementation phases.