Nguyen Tran Trung Thanh
Backend & Applied AI Engineer · Hanoi, Vietnam

Nguyen Tran Trung Thanh

Backend architecture, external integrations, and verifiable systems.

Working on backend systems, external integrations, and applied machine-learning services, with an emphasis on explicit boundaries, failure handling, and verifiable behavior.

Explore Verified WorkGitHub ProfileDirect Email
Invariant 01
Bounded Vendor Reads
Timeouts & limited retries
Invariant 02
Database Invariants
Idempotency & constraints
Invariant 03
Applied ML Decoupling
Estimation vs recommendation
Invariant 04
Disposable Testcontainers
Automated DB verification
01 // SELECTED WORK

Flagship Systems & Applied ML Projects

Detailed case studies documenting architecture, boundary constraints, and verification evidence.

Flagship Case Study

TRANSO AI Sales Bot

Role: Backend Architecture & External Integrations · Modular Monolith

Active Development (Milestone 7)

A modular Spring Boot backend engineered to ingest inbound Meta Messenger events, execute bounded KiotViet inventory reads, enforce vendor-neutral failure semantics, and persist durable employee-review work items with database-enforced invariants.

VERIFIED IMPLEMENTATION SCOPE
  • • Inbound Meta webhook HMAC-SHA256 signature verification
  • • Bounded KiotViet inventory reads with timeouts and limited retries
  • • Vendor-neutral integration failure translation
  • • Persisted employee-review work items with database invariants
  • • Automated testing against PostgreSQL via Testcontainers
EXPLICIT SYSTEM BOUNDARIES (NOT IN SCOPE)
  • • No message queues or Redis brokers
  • • No circuit breaker pattern (bounded reads via timeouts/retries)
  • • No completed outbound Meta Messenger message delivery
  • • No completed employee review UI/API or CRM sync
  • • No live production deployment or traffic claims
Java 21Spring BootPostgreSQLFlywayTestcontainersMeta WebhooksKiotViet API
Read Full Technical Case Study
Featured Applied AI Case Study

AI Fitness & Nutrition Recommendation System

Role: Applied ML & API Architecture · Containerized FastAPI Service

Containerized Open Source Service

A containerized FastAPI service decoupling physiological estimation (XGBoost for calories, Random Forest for BMI, Harris-Benedict for BMR/TDEE) from a two-stage hybrid recommender combining content-based scoring with a collaborative component over user-item interactions.

VERIFIED ARCHITECTURE & ML STACK
  • • XGBoost Regressor for active caloric expenditure estimation
  • • Random Forest Regressor for BMI with mathematical fallback
  • • Harris-Benedict formulas anchoring physiological BMR/TDEE
  • • Hybrid Recommender: Content-Based Cosine Similarity (0.6) + Collaborative Interaction Scoring (0.4)
  • • Endpoints for workouts (/recommend/workout) and meals (/recommend/meal)
RUNTIME & SYSTEM BOUNDARIES
  • • Multi-stage Dockerfile and Docker Compose orchestration
  • • Strict separation of biometric estimation from recommendation
  • • Non-clinical fitness tool (no medical or diagnostic claims)
  • • File-based/CSV feature catalog in V1 (no active relational DB)
PythonFastAPIXGBoostRandom ForestCollaborative Filteringscikit-learnDockerPydantic
Read Architecture Case StudySource Repository
02 // CAPABILITIES

Engineering Focus & Technical Competencies

Areas of hands-on practice, grounded directly in verified project implementations.

Backend Architecture

Building layered backend services with explicit domain isolation, typed contract validation, and strict boundary separation.

Java 21Spring BootFastAPIREST APIsWebhook IngestionModular Monoliths
External Integrations & Resilience

Integrating external commerce and messaging providers with defensive read bounds, idempotent event handling, and deterministic human-handoff rules.

Meta WebhooksKiotViet APIBounded ReadsTimeout & Limited RetriesVendor-Neutral Fault Isolation
Applied Machine Learning

Designing applied ML workflows that separate empirical biometric estimation from deterministic recommendation constraints and formulaic baselines.

PythonXGBoostRandom ForestCollaborative FilteringHybrid RecommendersFeature Pipelines
Persistence & Verification

Enforcing relational consistency and idempotency at the database level, with automated integration suites testing against containerized instances.

PostgreSQLFlywayDockerTestcontainersIntegration TestingDatabase Invariants
03 // CS FUNDAMENTALS

Operating Systems & Concurrency Architecture

Academic systems modeling verifying thread safety, process lifecycle state machines, and scheduling invariants.

CPU Scheduling & Concurrency Simulation

Role: Concurrency & Systems Programming · Java Systems Emulation

Systems Reference Architecture

A Java systems simulation demonstrating core operating system scheduling algorithms, starvation prevention through dynamic aging, and concurrent thread synchronization under simulated I/O blocking.

IMPLEMENTED SCHEDULING MECHANICS
  • • Round Robin scheduling with configurable time-slice quantum
  • • Preemptive and Non-preemptive Priority scheduling with dynamic aging algorithm to eliminate process starvation
  • • Process state machine: Ready → Running → Blocked on I/O → Terminated
  • • Thread-safe metric collectors calculating CPU utilization, wait time, and turnaround latency
JavaMultithreadingConcurrency UtilitiesData Structures
Source Repository
04 // EDUCATION

Academic Background

Formal computer science education and joint degree coursework.

Bachelor of Science in Computer Science

Troy University (USA) · Hanoi University of Science and Technology (HUST)

Joint Undergraduate Program
  • Rigorous foundations in Operating Systems, Concurrency, Algorithms, and Distributed Systems.
  • Hands-on systems implementation in Java, C++, and Python with strict focus on thread safety and data integrity.
05 // CONTACT

Get in Touch

Direct contact channels for technical inquiries and engineering roles.

Interested in discussing backend resilience, external vendor integration patterns, or applied machine learning pipelines? Reach out directly via email or inspect code repositories on GitHub.

trungthanh19092004@gmail.comgithub.com/trungthanh1909
Note: Resume CTA is omitted for V1 pending final document verification.