Mohiddin S

About Mohiddin S

I’m an AI/ML Engineer with 4+ years of experience designing, training, and deploying large-scale machine learning systems, LLM-powered applications, and cloud-native MLOps pipelines. I specialize in building end-to-end AI solutions using PyTorch, Transformers, Spark, and distributed architectures that deliver high accuracy, reliability, and real-time performance.My expertise spans the full lifecycle of AI development—data engineering, model training, optimization, deployment, monitoring, and automation. I’ve developed production-grade NLP systems, fine-tuned LLaMA/BERT models, built feature stores, engineered scalable microservices with Docker & Kubernetes, and delivered mission-critical ML workflows across healthcare and enterprise environments.At CVS Health, I work on AI-driven healthcare intelligence platforms—processing millions of clinical records using Spark, engineering ML features in Databricks, fine-tuning domain-specific LLMs, and deploying high-availability inference services on AWS EKS. My projects have improved prediction accuracy, accelerated clinical analytics, and supported safer, faster decision-making through RAG-based GPT applications.Previously at Vivma Software Inc, I engineered large-scale ML systems including personalized recommendation engines, reinforcement learning models, and enterprise-grade MLOps pipelines. I deployed high-throughput model services using Kubernetes, MLflow, and TorchServe, while implementing automated retraining, monitoring, and CI/CD workflows for rapid iteration and reliability.I’m driven by a passion for building scalable, efficient, and explainable AI systems that solve real-world problems. I enjoy combining deep technical knowledge with practical problem-solving, cross-functional collaboration, and continuous experimentation.

Experience

1. Ai Ml Engineer

   CVS Health

   Aug 2024 — Present

   Built automated data pipelines using Python, SQL, and Apache Airflow to process 2M patient records in compliance with HIPAA regulations, reducing data preprocessing time from 8 hours to 90 minutes and enabling faster AI-powered clinical analytics. • Developed real-time data processing systems with Apache Spark and PySpark to handle millions of patient records efficiently, cutting preprocessing time by 80% while maintaining HIPAA compliance standards for Machine Learning workflows. • Created a feature store using Databricks, Delta Lake, and MLflow that extracted 50+ clinical features from patient health records using SQL and Pandas, improving the accuracy of patient risk prediction ML models by 28%. • Fine-tuned LLaMA 2 and ClinicalBERT language models using PyTorch Lightning, LoRA, and Hugging Face Transformers on de-identified clinical notes, reducing GPU memory usage by 40% and lowering LLM training costs. • Deployed automated medical coding systems using BioBERT, PyTorch, Transformers, and spaCy that achieved 92% accuracy in real-time NLP processing, reducing manual coding workload by 30% across multiple departments. • Built patient readmission prediction models using XGBoost, LightGBM, and Random Forest with MLflow for experiment tracking and A/B testing in Agile sprints, helping reduce hospital readmissions by 11% organization-wide. • Developed a clinical decision support system using LangChain, Azure OpenAI GPT-4, Pinecone, and integrated with FDA knowledge bases, delivering responses in under 200ms through Retrieval-Augmented Generation (RAG) to help reduce adverse medical events. • Managed AI/ML model training and deployment infrastructure using PyTorch, Ray, Docker, Kubernetes on AWS EKS, and TorchServe with CI/CD automation through GitHub Actions, maintaining 99.5% system uptime in production.

Education

• The University of Texas at Arlington

  Master's degree, Information Technology

• PRIST University - India

  Bachelor's degree, Electrical, Electronics and Communications Engineering