Associate Machine Learning Engineer at CausalMind

Built an automated causal discovery system using Ollama LLMs and Neo4j, achieving 95.65% F1 and 91.67% recall on internal evaluation and improving graph-matching consistency. Added reliability guardrails (loop prevention + semantic validation), reducing failure cases by ~90% and stabilizing automated reasoning outputs. Developed a hybrid causal inference approach combining BIC scoring, PC algorithm, and LLM-based reasoning for complex graph structures. Containerized and deployed services using Docker on cloud environments to enable reproducible, scalable execution.