Duy Tien Nguyen
Backend Engineer
Backend engineer focused on system performance and correctness across IAM, caching, and service security. I own the auth service for the V2 rebuild of a six-service enterprise asset-management platform, from design and implementation through staging validation and maintenance. Improved permission-check performance by more than an order of magnitude in internal load tests.
Skills
Experience
SMART GENERATION OF DIGITAL
Ho Chi Minh City, Vietnam- Own the auth service for the V2 rebuild of a six-service enterprise asset-management platform, translating requirements into system design and implementation, reviewing architectural decisions with the tech lead, and maintaining the service through staging with frontend and QA.
- Anticipated permission evaluation as a scaling bottleneck and introduced a three-tier read path across Redis, Neo4j, and MongoDB, improving p99 latency by more than an order of magnitude under staging load tests.
- Moved hierarchical authorization checks to a Neo4j read model for assign/unassign validation and cycle detection, offloading recursive graph queries from the primary MongoDB model and reducing average hierarchy-check latency by roughly two-thirds in staging tests.
- Built event-driven invalidation for cached permission matrices using a Transactional Outbox and idempotent consumers, replacing TTL-only freshness with explicit permission-change events.
- Implemented the auth service's JWKS/key-rotation flow and file-encryption integration with the platform KMS, keeping application services decoupled from direct cryptographic key handling.
Selected Projects
Second Prize, University Research Competition • Prometheus, Grafana, Kafka, MinIO, Nginx, Docker
- Instrumented a backend stack with application and infrastructure telemetry, including Kafka consumer-group lag and system load, to identify message-processing bottlenecks that were not visible from application logs alone.
Advanced Algorithms — Individual Project • Python, Monte Carlo Simulation, Hypothesis Testing
- Benchmarked five bandit policies against round-robin and least-connections across stationary and drifting workloads using 150 Monte Carlo trajectories, with the best-performing policy reducing mean simulated request latency from 51.9 ms to 18.6 ms.
- Reported results that contradicted the original hypotheses, including policy-rank reversals under different tuning budgets and the failure of forgetting mechanisms to outperform alternatives at the tested drift densities.
Reinforcement Learning — Course Project • Python, Discrete CQL, PyTorch, Numba, Dynamic Programming
- Built a Safe-RL system combining Conservative Q-Learning with a hard constraint layer to optimize transaction timing under gas-price volatility while maintaining zero deadline-constraint violations in the evaluated experiments.
Anchored IPFS content hashes of donation evidence on-chain, making the audit trail tamper-evident rather than trust-based.
Education
HO CHI MINH CITY UNIVERSITY OF TRANSPORT
Expected Dec 2026Bachelor of Science in Data Science — GPA 3.86/4.0
Honors: Eureka semi-finalist (2025) • Merit Scholarship, Top 1%
Certifications: Google Cybersecurity Professional (2025) • IBM DevOps & Software Engineering (2026)