Ali Maksudi

Hupo (Contract)

Insurance System Migration (European Financial Client)

• Applied unsupervised learning (K-Means, hierarchical clustering) to segment and consolidate 200+ legacy insurance products, enabling streamlined product portfolios and improved maintainability.
• Built end-to-end data pipelines for document ingestion (OCR + parsing), transforming unstructured policy documents into structured features used in downstream ML workflows.
• Designed scalable data architectures for AI systems, improving pipeline reliability and reducing processing latency by ~30%.
• Provided technical leadership across AI initiatives, influencing system design decisions for production-grade deployments across multiple teams.

SQE

Enterprise AI Systems (LLM, RAG, Multi-Agent)

• Architected and deployed an HR AI assistant handling 10k+ queries/month, automating ~45% of employee inquiries and significantly reducing HR operational workload.
• Built a real-time investment advisory platform integrating external financial APIs, improving analysis speed by ~2x and enabling automated reporting for stock performance.
• Developed an LLM-powered invoice processing system (OCR + entity extraction), reducing manual data entry time by ~60% and improving compliance processing speed.
• Orchestrated multi-agent LLM workflows using LangGraph, enabling structured reasoning and reliable tool usage across distributed AI services.
• Optimized RAG pipelines (LangChain, LlamaIndex) through advanced chunking, metadata filtering, and re-ranking, improving retrieval accuracy by ~25%.
• Scaled production AI services using Docker, FastAPI, and CI/CD pipelines on Azure, achieving high system reliability and faster deployment cycles.

Sinarmas Mining

Credit Score Development

• Led end-to-end development of credit scoring models deployed across 12+ financial institutions, improving lending decision accuracy by ~20% and reducing default rates by ~12%.
• Built and deployed ML pipelines on Alibaba Cloud and Cloudera, ensuring model scalability, monitoring, and regulatory compliance.
• Designed feature engineering frameworks for large-scale financial datasets, improving model robustness and interpretability.
• Implemented model monitoring and governance systems, ensuring stability and compliance across production environments.

JD.ID

• Developed hybrid fraud detection systems (ML + rule-based), achieving 95% detection accuracy and reducing fraud-related losses by ~30%.
• Built real-time ETL pipelines and dashboards for fraud monitoring, improving operational visibility and response time.

Mercu Buana Uni

• Taught Introduction to Algorithms and Object-Oriented Programming (Python) to undergraduate students.