Manohar Lakavath

• Improved forecasting accuracy by 18% using Python, Scikit-learn, and Pandas, enabling faster enterprise decisions across pipelines while reducing manual errors and operational inefficiencies. • Fine-tuned CNN and Transformer models for NLP and computer vision tasks, achieving 92% validation accuracy and reducing inference latency 20%, increasing production efficiency for multiple business units. • Built end-to-end MLOps pipelines with Docker, Flask/FastAPI, and CI/CD, cutting deployment time 40% while ensuring scalability, reproducibility, and continuous monitoring of production-ready AI/ML solutions. • Implemented LLM-based solutions for text summarization, sentiment analysis, and question-answering, reducing task completion time 30% and providing actionable insights for enterprise clients across departments. • Optimized AWS SageMaker workloads and leveraged spot instances, reducing cloud infrastructure costs 25% while maintaining high availability, reliability, and performance of large-scale machine learning pipelines. • Designed real-time monitoring and alerting frameworks, decreasing production failures 20% and ensuring early anomaly detection, continuous performance, and proactive operational risk mitigation. • Integrated bias detection, fairness metrics, and privacy-preserving techniques, achieving 100% compliance with internal governance, GDPR, HIPAA, and responsible AI standards across all deployed models. • Developed interactive dashboards with Power BI and Tableau, visualizing model predictions, KPIs, and trends, improving executive decision-making speed 15% and driving data-informed business strategies.

•Built supervised and unsupervised ML models using Python, Scikit-learn, and Pandas, improving predictive accuracy 15–20% across retail, finance, and healthcare client projects. •Trained and deployed deep learning models including CNNs, RNNs, and Transformers on AWS SageMaker and Azure ML, handling datasets exceeding 10M records per project for production use. •Designed automated data preprocessing and feature engineering pipelines using Python and Pandas, reducing manual preparation time 35% while improving data quality and model input efficiency. • Conducted hyperparameter tuning, model evaluation, and cross-validation using TensorFlow and PyTorch to improve predictive performance and robustness of production-ready AI/ML models. • Standardized feature engineering and data cleaning workflows, improving model performance 18% and ensuring robust predictive insights across multiple industries. • Delivered Tableau and Power BI dashboards integrating model outputs, reducing reporting cycles 30% and enabling executive stakeholders to make faster, data-driven business decisions. • Automated experimentation and model training workflows using Azure ML and Google Colab, cutting prototyping and testing cycles 35% while accelerating deployment of AI/ML solutions. • Applied containerized ML deployments using Docker, increasing model production success rates 28% and ensuring high-performance delivery for client-facing applications.