Pham Minh Long

Pham Minh Long

✨ AI Solutions Architect  ·  Researcher @ ACM Lab, NYCU

"BE YOUR BEST SELF!"
🤖 LLM / AI Agent 🔬 Multimodal Learning ☁️ MLOps / Cloud 🔐 AI Agent Security ⚡ Distributed Systems

About Me

📋 Basic Info

  • Name: Pham Minh Long
  • Phone: (+84) 036 762 3811
  • Email: phamlong12082001@gmail.com
  • Languages: Vietnamese, English (IELTS 7.5)
  • Location: Ho Chi Minh City, Vietnam
  • Lab: ACM Lab, NYCU
  • Research: LLM · AI Agent Security · Multimodal Learning · Distributed Systems

🎓 Education

National Yang Ming Chiao Tung University

Master's student · ACM Lab

In Progress
2026 – present

HCMC University of Technology and Education

B.Eng. Computer Engineering · Valedictorian

GPA 3.25/4.00 · Top 1/60
Aug 2019 – Sep 2023

🛠️ Technical Skills

Languages

PythonC/C++ JavaNextJS

Deep Learning

PyTorchTensorFlow OpenCVONNX TensorRTTFLite WandB

LLM / ADK

LangChainLangGraph LlamaIndexCrewAI vLLMOllama

MLOps / DevOps

DockerKubernetes AirflowMLflow ArgoCDGrafana PrometheusGitLab-CI

Cloud & Infrastructure

AWSGCP HAProxyPortainer

Databases

FaissQdrant ElasticSearchChroma MySQLMongoDB RedisBigQuery

Knowledges

Graph RAGMulti-Agent LoRA / RLHF / GRPODPO System DesignLoad Balancing Event-driven

ML / Data

PandasApache Spark Scikit-learnXGBoost PolarsNumpy

Professional Experience

🚀 3+ years of AI/ML engineering across fintech, cloud, and software product companies.

MoMo

Middle AI Engineer

Jun 2025 – Mar 2026
🏆 Top 2 Innovation Product Prize at MoMo
  • Enhanced Mimir Chatbot Platform — automated report generation, SQL/Chart/Insights using LangGraph + Deep Research pattern within 2 months
  • Implemented Jira-automated assignment via LangGraph for MoMo Data Scientists and employees
  • Integrated MCP Server and optimized historical conversation usage; weekly monitoring via Grafana + LangFuse
  • Performance: Knowledge Retention 0.98 · Turn Relevancy 0.74 · LLM-based SQL Accuracy 0.64 · P90 75.2s
LangGraphLangChain FastAPIRedis BigQueryQdrant Azure OpenAI

Nexcel Solutions

ML Engineer

Aug 2023 – Jun 2025
⭐ 2× ME+ · 1× EE Performance Ratings
  • Built Agentic RAG baseline + Web UI (Nextjs) for Betting Knowledge chatbot within 2 weeks; fine-tuned Llama-3.2-7B & Qwen2.5-7B via LoRA/PEFT
  • Designed multi-agent Report Entity Recognition system with Telegram integration
  • Analyzed & adjusted XGBoost loss function for Customer Time-Series Prediction (0.35 TP · 0.92 TN)
  • Pioneered CI/CD pipeline with multi-stage Dockerfiles, Airflow scheduling, Portainer management
  • Webiometric Data Streaming: Kafka → MySQL (SP + schema design) from Taiwan producer
LangChainOllama vLLMFastAPI AirflowQdrant HAProxy

Unicloud Group

AI Engineer

Feb 2022 – Aug 2023
  • Designed and improved Unicloud eKYC ID Card Verification system — YOLOv5/v7/v8 for 4-corner detection, PaddleOCR for information extraction
  • Added orientation correction without models to enhance inference time and accuracy
  • Performance: 64 fps · 0.87 mAP@50 · 0.94 mAP50-95 · 0.89 F1-score · 1.67 MB Docker image
YOLOv8PaddleOCR FastAPIDocker GCP
🔬

UTE-AI Lab

Research Assistant

May 2021 – Dec 2023
  • Participated in numerous competitions and conferences in HCMC
  • Tracked state-of-the-art AI research trends under PhD. Tran Vu Hoang

Research & Projects

🤖 Mimir Chatbot Platform

Jun 2025 – Mar 2026

MoMo · Middle AI Engineer · Team Size: 3

Enhanced internal chatbot with LangGraph Deep Research pattern for automated SQL, Chart, and Insights generation. Integrated MCP Server and Jira automation.

KR: 0.98 · TR: 0.74 · SQL: 0.64 · Chart: 0.64 · P90: 75.2s
LangGraphFastAPI Azure OpenAIBigQuery

🏷️ Report Entity Recognition

Feb – Jun 2025

Nexcel Solutions · Junior ML Engineer · Team Size: 2

Multi-agent system for multi-report generation across departments. Fine-tuned Qwen2.5-7B with bidirectional transformer for entity & span recognition.

Qwen2.5-7BOllama HuggingFaceHAProxy
🤗 HuggingFace

💬 Domain Chatbot – Betting Knowledge

Dec 2024 – Jun 2025

Nexcel Solutions · Junior ML Engineer · Team Size: 2

Agentic RAG baseline + Web UI within 1 week. Fine-tuned Llama-3.2-7B & Qwen2.5-7B via LoRA/PEFT on internal betting knowledge.

Precision: 0.74 · Faithfulness: 0.89 · Relevancy: 0.67
Llama-3.2-7BvLLM Next.jsQdrant

📈 Customer Performance Prediction

Jun – Dec 2024

Nexcel Solutions · Junior ML Engineer · Team Size: 2

Time-series RFM model for Taiwan punters' daily performance. Adapted XGBoost loss function for company winloss. Applied active learning & unsupervised domain adaptation.

TP: 0.35 · TN: 0.92
XGBoostK-Means RFM ModelUMAP

🎥 Text-Video Retrieval

Jul – Oct 2023

HCMC AI Challenge 2023 · Leader · Team Size: 5

Multi-modal search engine: top-k retrieval from video/image/text using CLIP, ASR, Image Captioning, OCR for feature extraction and cosine similarity.

CLIPBERT FaissElasticsearch FastAPI

📱 Smart Menu Application

Feb – Jun 2023

Graduation Project · Researcher · Team Size: 1

Mobile app: Menu Scanner (PaddleOCR + C++ core), Machine Translation (RetNet/Transformer), Recommendation (NeuMF + Collaborative Filtering), RASA chatbot.

JavaPaddleOCR BERTGCP

🪪 ID Card Verification – eKYC

Feb 2022 – Aug 2023

Unicloud Group · AI Engineer · Team Size: 5

Improved eKYC system with YOLOv5/v7/v8 for 4-corner detection, perspective transform for ROI extraction, PaddleOCR for text extraction.

64 fps · mAP@50: 0.87 · F1: 0.89 · Image: 1.67 MB
YOLOv8PaddleOCR Docker

🚗 Autonomous Car – CV & DL

Sep – Nov 2022

UIT Racing Car 2023 · Leader · Team Size: 4

UNET 3+ / BiseNet lane segmentation with PID controller. YOLOv4-tiny/v7/v8 traffic sign detection. Simulated on Unity, embedded on Jetson Nano.

YOLOv8UNET 3+ Jetson NanoUnity

Awards & Honors

🥈

Top 2 Innovation Product Prize

MoMo · 2025 — Mimir Chatbot Platform enhancement for Deep Research & automated reporting

🎓

Valedictorian – Computer Engineering

HCMUTE K19 · 2023 — Top 1/60, GPA 3.25/4.00 (8.21/10.00)

⭐

EE Performance Rating

Nexcel Solutions — Exceeding Expectations, alongside 2× ME+ ratings

🏅 Certifications

Solutions Architect Associate

Machine Learning Engineer Associate

SysOps Administrator (CloudOps) Associate

Deep Learning with PyTorch: Siamese Network

Coursera

Deep Learning with PyTorch: GAN

Coursera

Deep Learning with PyTorch: Neural Style Transfer

Coursera

Deep Learning with PyTorch: Object Localization

Coursera

Deep Learning with PyTorch: GradCAM

Coursera

Convolutional Neural Networks

Coursera

Facial Expression Recognition with PyTorch

Coursera

Aerial Image Segmentation with PyTorch

Coursera

Deep Learning with PyTorch: Image Segmentation

Coursera

Publications

Research in Progress – ACM Lab, NYCU

Currently conducting research at ACM Lab, National Yang Ming Chiao Tung University (NYCU) on topics including LLM, AI Agent Security, Multimodal Learning, and Distributed Systems.

ACM Lab, NYCU · 2026 – present

Full list of publications and preprints available on Google Scholar.

View Google Scholar 🤗 HuggingFace Models