Shuja Ur Rahman Mohammad

Ai Ml Engineer @ServiceNow

US

Orchestrated instruction-tuned LLMs with LoRA/QLoRA, improving accuracy by about 20% while reducing computational overhead by nearly20%, enhancing production performance.• Devised retrieval-augmented pipelines with LangChain and Pinecone, cutting query latency by around one-quarter and strengthening contextualprecision in enterprise-scale deployments.• Streamlined end-to-end model lifecycle with MLflow, Airflow, and Kubernetes, shortening release cycles by roughly 40%, ensuring reliableproduction environments.• Applied Bayesian structural models and Isolation Forest to healthcare data, lowering false alarms by about one-quarter, supporting regulatorycompliance.• Refined inference with ONNX runtime optimizations, reducing response delays by about one-third, enabling faster real-time interactions.• Directed knowledge transfer sessions on RAG, MLOps, and multi-agent coordination, improving team productivity by around 20%.