Satish Beladiya

Integrated large language models like GPT, LLaMA, and Mistral into real-world pipelines for classification, extraction, summarization, and content generation with high accuracy on imbalanced datasets. Automated internal processes using multi-agent AI systems via MCP and LangChain, boosting operational efficiency and client conversion. Built RAG-based agentic workflows with LangChain, OpenAI APIs, and ChromaDB, increasing user engagement. Fine-tuned and deployed BERT and GPT-based models achieving a 33% F1-score improvement. Utilized classical ML techniques to establish baselines and complement deep learning solutions. Crafted advanced prompts to reduce hallucination and improve reasoning. Implemented model monitoring, drift detection, and A/B testing for stable performance. Deployed scalable pipelines on AWS/Azure with Docker and Kubernetes. Collaborated cross-functionally to standardize AI workflows and presented results to stakeholders.

Trained and fine-tuned YOLOv7, SSD, and CenterNet object detection models for waste classification, achieving high precision and 60% higher FPS. Designed and deployed an active learning pipeline, boosting F1-score by 15 points and reducing annotation costs by 50%. Delivered OscarPocket, a client chatbot for waste regulation queries, enhancing compliance and satisfaction. Optimized models using TensorRT and ONNX for deployment on NVIDIA Jetson devices in real-time constrained environments. Upgraded hardware and streamlined supply chain to reduce operational costs by 50%, and optimized storage cutting costs by 40%. Reduced training cycles from 7 to 3 days using high-efficiency GPUs, saving 30% in costs. Built cloud pipelines with automated monitoring and CI/CD enabling continuous performance.

Led a team of 3 engineers managing six projects with clients across multiple time zones, ensuring on-time delivery. Applied Python and machine learning to optimize manufacturing processes, reducing production time by 20%. Coordinated with internal and external stakeholders to meet deadlines and exceed client expectations. Demonstrated leadership in managing complex engineering projects under tight schedules and evolving requirements.