Siva Manikanta Lakumarapu

About

AI/LLM engineer with production experience building secure, scalable compliance-focused generative AI systems (GPT-3/4, BERT) including RAG over internal regulatory document bases. Has delivered end-to-end pipelines on AWS with PySpark/Airflow/Kubernetes/FastAPI, emphasizing privacy controls, monitoring, and iterative evaluation (A/B testing). Also partnered closely with bank compliance officers using prototypes to refine NLP summarization/classification and reduce document review time.

Experience

Education

Key Strengths

• Built and deployed production generative AI system for regulatory/compliance document summarization and clause extraction
• Designed end-to-end LLM pipeline (ingestion, preprocessing, summarization, RAG retrieval) on AWS with scalable microservices
• Handled inconsistent/unexpected document formats via custom PySpark ETL normalization/cleaning
• Implemented security controls for sensitive documents (encryption in transit/at rest, IAM, authenticated FastAPI endpoints)
• Reduced hallucinations and improved accuracy via prompt engineering, few-shot examples, and continuous A/B testing
• Production orchestration experience with Airflow (DAGs for ETL/ML workflows) and Kubernetes (Helm, autoscaling, troubleshooting)
• Reliability practices: unit/integration testing with PyTest, CI/CD automation, monitoring with Grafana/Prometheus
• Effective collaboration with non-technical compliance stakeholders using discovery sessions and Streamlit prototypes to iterate on false positives/negatives