Minh Vu

• FinTech cash flow forecasting platform. Built self-serve CSV/Excel data onboarding for uploads of up to 10,000 rows, replacing a multi-hour manual import the principal engineer repeated for every customer data delivery; column mapping uses deterministic rules first and an LLM as a fallback, and transactions land in an append-only PostgreSQL ledger.
• Built transaction categorization: a rule-based categorizer over transaction descriptions and payer/payee that the principal engineer used for imports, then an LLM pipeline that categorizes transactions automatically.
• In React, created a streaming AI chat window that shows the assistant's status updates as well as its answer, and a paginated transaction table.

• AI-powered invoice financing platform (contributed to $2M funding round). First engineer on the AI team (grew to 4), shipping v1 with 2 senior Grapple engineers; owned the end-to-end assessment workflow for invoice financing, from document upload to the interactive results dashboard, running on AWS (ECS, Lambda, SQS).
• Reduced operational workload by 250+ hours/week, cut p95 assessment time for 100-page contracts from 20+ min to under 2 min. Cut per-assessment cost by 75% ($2 → $0.50) via Redis and provider-side caching (80–90% hit rate).

• Elastic Stack consultancy. Wrote Logstash parsing scripts for 5 client projects (a Vietnamese bank and 4 in Singapore), extracting structured fields from raw logs; also wrote Filebeat log-shipping configurations.
• On-site at that bank, built 7 Kibana visualizations tracking latency, uptime and error counts across web, mobile and internal systems, plus a Kibana Canvas status board with threshold-based coloring.

University of Passau, Germany

Focusing on compilers and language implementation — IR design and optimization passes, static program analysis, runtime systems, and parallel/high-performance computing.

FPT University, Vietnam

GPA: 8.94/10 · Full Scholarship · Top 10/900+ students