Nguyen Hoang Long

About Me

I am an AI Researcher and Software Engineer with 4 years of combat experience in the tech trenches. I don't just code; I engage in deep research to solve problems that others deem impossible. My superpower is Nonlinear Learning Speed—I can dive into a completely new domain and master it in record time. I describe myself as Creative, Hardcore, and Competitive. Whether it's competitive programming, security research, or building AGI-level architectures, I play to win and innovate.

Fueled by Hardcore Music & Late Night Coding

Technical Arsenal

Languages

Python (Expert)
C/C++ (Expert)
TypeScript (Advanced)

AI & Data Science

PyTorch (Expert)
VectorDB (Expert)
GraphDB (Advanced)

Specialty

Metacognitive Architectures
Drug Repurposing
R&D

Backend & Frameworks

FastAPI (Master)
Django (Expert)
NestJS (Expert)
Next.js (Advanced)

Infrastructure & DevOps

Git
CI/CD
Docker
AWS
K6 (Performance Testing)

Database

SQL (PostgreSQL/MariaDB)
Redis
Kafka (Event Streaming)
MongoDB
MinIO

Featured Projects (The Highlights)

BioMind-DR: Biomedical Metacognitive Intelligence

The Magnum Opus. A system that mimics the human brain to cure diseases.

Role: Project Leader & Lead Researcher (1 Year)

The Problem: Traditional drug repurposing is slow and disconnected. We needed a system that doesn't just "search" but "thinks" and verifies hypotheses like a scientist.

The Solution: Built a Leader-Expert-Council Multi-Agent System based on Metacognitive Memory Architecture (simulating Hippocampus & Neocortex).

Fast System (Hippocampus): L0 Working Memory (Redis) & L1 Episodic Buffer (Kafka) for real-time context and event streaming.
Slow System (Neocortex): L2 Knowledge Core powered by KuzuDB.
Bi-temporal Graph: Tracks validity of knowledge over history.
Evidence Graph: Isolates conflicting data for the Council Agent to adjudicate.
Memory Consolidation: L3 Strategy (MariaDB) & L4 Long-term Storage (MinIO) allow the system to learn from past inferences.

Impact & Result:

Outperformed SOTA models: Achieved 48.2% on HLE and 94.7% on MRCR v2, beating Gemini 3.0 Pro and GPT-5.1 in biomedical reasoning tasks.
Integrated a virtual dry-lab for hypothesis verification.

Tech Stack: Python, PyTorch, KuzuDB, Kafka, Redis, MariaDB, MinIO.

Link: Live Demo

PulsarPath: Next-Gen Personalized Education AI

Redefining how we learn.

Role: Solo Developer (3 Months - Closed Beta)

Concept: An AI-driven platform that analyzes individual thinking patterns to generate hyper-personalized learning roadmaps. It optimizes for the user's specific cognitive style to maximize learning efficiency.

Tech Stack: Next.js, NestJS, AI Analytics Engine.

Link: pulsarpath.dev

Education & Achievements

Le Hong Phong High School

2023 - 2026

Focus: Computer Science & AI.

Honors & Awards

Top 2% Competitive Programming in Vietnam (Age Group).
Top 10% National Science & Engineering Fair in Vietnam.
Security Research: Reported critical vulnerabilities to EdTech platforms (Edubit/Unica, Eduquiz) affecting ~700k users.
Open Source: Contributor to Letta, VNOI and Hetio.