Vinaykumar

Architected and developed production-ready AI agent system for Aircraft maintenance troubleshooting using LangGraph, OpenAI and Weaviate vector search. Designed master-worker architecture with hierarchical graph workflows for multi-document maintenance procedures including FIM, AMM, MMEL, and BITE. Built hybrid document processing framework combining LlamaParse PDF extraction with Neo4j graph database, achieving 10,000-40,000x performance improvement over LLM-based parsing. Implemented advanced state management utilizing dual context windows with sliding window patterns and Memento pattern for comprehensive audit trails. Developed scalable streaming API architecture with FastAPI and Redis persistence for multi-turn conversations with state restoration capabilities. Created specialized document processors for maintenance manuals supporting configurable extraction pipelines for FIM fault isolation, AMM tasks, and MMEL limitations. Engineered performance optimizations reducing walkthrough initialization from 40+ seconds to sub-millisecond response times through structured data extraction. Integrated multiple AI/ML technologies including LangChain, LlamaIndex, Weaviate vector database, Neo4j, and OpenAI API for robust data modeling. Developed scalable software solutions using Agile methodologies.

Mentored Infosys Springboard interns across six batches in AI/ML and data visualization projects. Facilitated skill development in emerging technologies through targeted guidance. Applied Agile methodology to support interns throughout project lifecycles.

Working directly with the stakeholders for their requirements, helped them ship their AI solutions by addressing their maintenance, advisory and technical needs. Prompt Engineering with GPT-3.5, GPT 4 models. Finetuning the base models of GPT-3.5, GPT 4, GenAI models. Worked with pre-trained models from Open AI and Microsoft Research available in Model catalog for fine-tuning and inference tasks for customer use cases. Discovered and reported potential bugs in the system. Rapidly Integrated into the Data Science Division quickly tackling problems on tickets for Cognitive services, Open AI and solving issues as well as boosted the performance of solutions. ML Ops integrates machine learning models into CI/CD workflows, enabling automated testing, validation, and deployment in production environments.

Prototyped machine learning applications and quickly determined application viability. Image detection and annotating the images. Transformed raw data to confirm to assumptions of machine learning algorithm. Incorporated Image and Video data to understand and translate complex business problems into machine learning problems. Data collection and labeling data for better performance of algorithm. Data analyzing and creating reports according to the analysis. Following SOPs according to business understanding. Data gathering using SQL tools to better maintain productivity of delivery system. Responsible for the foundational labelling functions in the context of Alexa's performance improvement using a Large Language Model. Worked with AWS Sagemaker studio for building models. Working in a diverse team with a shared vision, I contribute through independent thinking, problem-solving, and collaboration.

Used ML Tools for Google Ad-words to manage advertising stakeholders. Prioritized tasks to meet tight deadlines, pitching in to assist others with project duties. Utilized customer feedback to improve product offerings.

Provided services like Cab advertising, Outdoor Advertising, Web designing, branding and Social media marketing.