PIJUSH PAL

Designed and trained AI/ML models for referral matching, patient triage, and intelligent appointment recommendations in clinical settings. Developed NLP pipelines for SOAP note summarization and real-time patient communication, enhancing clinical documentation efficiency. Built and orchestrated multi-agent systems using AutoGen, llama-Index and CrewAI to enable autonomous, context-aware healthcare

Built enterprise-grade RAG systems with Neo4j to boost contextual search speed by 60%. Developed multi-agent LLM workflows using the ReAct framework for real-time tool-use and knowledge retrieval. Deployed multimodal GenAI pipelines (text + image, TTS + ASR) using Redis caching and optimized prompt execution, reducing latency by 45%.

Enhanced scalability of data processing by 30% with machine learning pipelines, contributing to more efficient deployments of AI projects. Enabled quicker and more accurate decision-making through performance optimization and A/B testing, improving project deployment competently. Successfully deployed advanced predictive models on Amazon SageMaker, refining the software development process for rob

Reduced evaluation time from 20 minutes to 12 minutes by engineering a decision-making platform for global postgraduate admissions. Led a team of 15 members to design custom ML algorithms, enhancing performance metrics by 20%, which improved the accuracy and reliability of the system. Achieved 75-80% accuracy in admissions procedures by developing a Streamlit application and predictive model.

Activities and societies: TA in ML Course Curriculum

Completed a Master's degree in Data Science. Served as a Teaching Assistant for an ML Course Curriculum.