tahutu/cv
Lead AI Engineer Updated September 2026

Thuc Tran

Lead AI Engineer
GenAI · Engineering · Product Mindset
10,000+ Daily Real-Time Calls (<2s E2E Latency)
Top 3 Global Rank @ Interspeech 2026 (MLC-SLM)
Top 56 Global Rank @ NIST FRVT 1:1 Benchmark
1st Place VLSP 2025 Challenge (Speech Quality)
iBeta 1 & 2 FIDO Alliance Presentation-Attack Compliance
Lead AI Engineer with a strong product mindset, bridging cutting-edge AI research and enterprise production scale. Extensive track record architecting real-time multi-agent conversational systems, multilingual speech language models (SLM), and high-throughput computer vision platforms. Deep hands-on experience across the entire model lifecycle: from training and fine-tuning to sub-second inference optimization, hardware acceleration on NVIDIA H100s, and resilient MLOps architecture.

Experience

Cake Digital Bank

Ho Chi Minh City, Vietnam · 2 yrs 5 mos
Lead AI Engineer June 2026 – Present
  • High-Volume Agentic AI: Deployed real-time multi-agent systems for enterprise call centers, handling 10,000+ daily calls with <2s end-to-end latency.
  • State-of-the-Art Voice Models: Engineered an advanced multilingual speech language model (SLM), achieving 3rd place globally at Interspeech 2026.
  • Enterprise LLM Architecture: Spearheaded the deployment of secure generative AI pipelines, bridging cutting-edge ML research with enterprise scale and security governance.
Senior AI Engineer May 2024 – May 2026 · 2 yrs 1 mo
  • Enterprise Automation: Deployed enterprise-wide generative AI applications to automate and optimize front, middle, and back-office banking operations.
  • Compliance & Benchmarks: Secured iBeta Level 1 and Level 2 compliance certifications for passive liveness verification, and won 1st place at the VLSP 2025 challenge.

VNG Digital Business – AI Lab

Ho Chi Minh City, Vietnam · 1 yr 1 mo
AI Engineer April 2023 – April 2024
  • Generative AI & Diffusion Models: Led the end-to-end development of large-scale text-to-image systems based on Stable Diffusion on NVIDIA H100 GPUs, cutting overall inference latency by >40% via dynamic batching and asynchronous pipelines.
  • High-Throughput Serving & MLOps: Architected scalable Python backend services and production monitoring pipelines (Triton Inference Server, Prometheus, Grafana, Kibana) to serve generative AI workloads with enterprise stability.

VNG TrueID

Ho Chi Minh City, Vietnam · 1 yr 9 mos
AI Engineer August 2021 – April 2023
  • World-Class Biometrics & CV: Engineered the TrueID facial recognition and ID verification system, achieving a top-56 global ranking on the prestigious NIST FRVT 1:1 benchmark on its inaugural submission.
  • Production Biometrics Architecture: Developed low-latency computer vision microservices for facial detection, alignment, feature extraction, and anti-spoofing verification at national banking scale.

VNG Cloud

Vietnam · 10 mos
Associate Software Engineer (During university) November 2020 – August 2021
  • Gained early-career backend engineering experience alongside academic studies, building foundational backend systems, data pipelines, and microservices on GCP and VNG Cloud.

Key AI Systems & Production Work

Lead Architect 2024 – Present

Real-Time Multi-Agent Voice Systems

High-volume voice bot platform serving 10,000+ daily call-center calls. Orchestrates TTS, STT, VAD, and LLM backends with sub-2-second end-to-end streaming latency.

Voice Agents Streaming TTS/STT LLM Orchestration Low-Latency
Core Researcher 2026

Multilingual Speech Language Model (SLM)

Engineered advanced conversational speech diarization and recognition models, ranking 3rd globally at Interspeech 2026 (2nd MLC-SLM Challenge).

Speech LM Diarization Interspeech 2026 PyTorch
Tech Lead / AI PIC 2024 – 2025

eKYC Liveness & Fraud Prevention

Main driver for FIDO Alliance certification, achieving iBeta Level 1 & Level 2 passive presentation-attack detection (PAD) compliance and winning 1st place in VLSP 2025.

iBeta Level 1 & 2 Anti-Spoofing eKYC VLSP 2025 1st
Tech Lead 2023 – 2024

High-Throughput Stable Diffusion Platform

Commercial generative image platform on NVIDIA H100 GPUs. Decreased latency by >40% through dynamic batching, asynchronous queuing, and Triton Inference Server.

Diffusion NVIDIA H100 Triton TensorRT
CV Researcher 2021 – 2023

TrueID Biometrics & Face Recognition

Trained high-accuracy face embedding models serving millions of authentications. Placed Top 56 globally on the prestigious NIST FRVT 1:1 benchmark on first submission.

NIST FRVT Top 56 Face Recognition Loss Optimization PyTorch

Technical Capabilities

Generative AI & Voice
Multi-Agent Systems Speech Language Models (SLM) Real-time Voice Bots TTS / STT / VAD LLM Orchestration Prompt Engineering RAG Architectures
Machine Learning & Computer Vision
PyTorch TensorFlow ONNX TensorRT Stable Diffusion Face Recognition Anti-spoofing (iBeta 1 & 2)
MLOps & Production Serving
Triton Inference Server NVIDIA H100 / A100 Stack Docker Prometheus & Grafana Kibana & Elasticsearch GCP & AWS
Systems & Engineering
Python C++ gRPC Microservices Redis MySQL RabbitMQ

Awards & Honors

3rd Place Global June 2026
Interspeech 2026 (2nd MLC-SLM)
Task I: Multilingual Conversational Speech Diarization and Recognition.
1st Place October 2025
VLSP 2025 Challenge
Track 1: Speech Quality Assessment (Awarded ACL Anthology paper).
Top 56 Global 2022
NIST FRVT 1:1 Verification Benchmark
Global evaluation of TrueID face recognition on inaugural submission.

Education & Credentials

Ho Chi Minh City University of Technology (HCMUT - Bach Khoa)

Bachelor of Science in Computer Science · The Honors Program
GPA: 8.5 / 10
2017 – 2021

Looking for a complete PDF version?

Download the formatted, two-page LaTeX resume suitable for application tracking systems and offline review.