Aseem Kumar Sharma

Architected and led the end-to-end development of a generative AI application automating resume creation for end-users. Engineered a sophisticated RAG pipeline by fine-tuning LLMs like Mistral 7B and leveraging LangChain. Designed scalable backend architecture with vector databases such as pgvector, Qdrant, and Pinecone, including a novel hybrid Vector/Graph DB solution for improved data retrieval performance. Established a full CI/CD pipeline using Amazon CDK, automating deployment and increasing development velocity significantly.

Designed and deployed a production-scale hybrid recommendation engine on AWS, boosting targeted user engagement for a media streaming platform. Developed and trained reinforcement learning agents for complex gaming environments like Doom and Mario using curriculum and sequential learning. Engineered diverse ML solutions, including generative AI models for content synthesis and YOLO-based object detection systems to improve accuracy in security and industrial use cases.

Translated complex business requirements for major international clients such as Sony PlayStation, Goosehead, Moog Insurance, and Idexx Labs into high-impact data solutions. Engineered a recommendation engine for Sony's PlayStation digital ad-sales platform using sentiment analysis and customer history. Led unified data infrastructure creation for IDEXX Labs by migrating multiple sources into a single source of truth. Built and deployed customer attrition and opportunity conversion models using Salesforce Einstein, Spark, AWS SageMaker, and integrated various data platforms to improve client retention and optimize sales forecasting.