Computer Science Student
Open to Internship & Freelance

Building software that solves real-world problems.

Computer Science student focused on full-stack development, AI systems, RAG pipelines, and modern software engineering.

developer.json
$ cat developer.json
{
"name": "Nguyễn Tuấn Khôi",
"alias": "callm3tk",
"stack": ["RAG", "Laravel", "React"],
"focus": "LLMs & Scalability"
}
$ _
1,290+PDFs Processed
11M+Tokens Indexed
15K+Chunks Generated
RealtimeApplications Built
AISystems Designed

Overview

An architectural overview of my development philosophy, workflows, and technical stack.

Engineering Philosophy

I believe in building software that delivers immediate, tangible value. Rather than chasing theoretical abstractions, my focus is on product thinking, problem solving, and writing practical, maintainable code. I bridge the gap between engineering execution and product design, striving to solve real-world problems with solid architecture and artificial intelligence.

Current Focus

  • Retrieval-Augmented Generation (RAG)
  • AI applications & LLM Orchestration
  • Full Stack Web Development
  • Real-time Systems (WebSockets & Queues)
  • Developer Experience (DX)

Open Source

Committed to learning in public and shipping high-quality, clean codebases. Active in developer ecosystems.

PRs MergedGit Commits
github.com/callm3tk

AI Pipeline

High-performance retrieval pipeline mapping dense vector indexing to Gemini model responses.

Query
Embedding (BGE-M3)
Qdrant DB
Retrieval
Gemini LLM
Response

Realtime Architecture

Low-latency topology orchestrating Docker, Nginx, Redis caching, and Laravel queue workers.

Docker
Nginx
Laravel
Redis Cache
Reverb
Client
Featured Platform

UE Connect

Production

A comprehensive university social network and academic mentoring platform integrating event-driven communication feeds with a self-hosted semantic advising AI pipeline.

LaravelPHPMySQLPython

Languages

Python
TypeScript
JavaScript
PHP
C++

GitHub Activity

User:callm3tk
Status:Active
Repos:Public
View Activity Grid

Featured Projects

Production-grade systems and AI-powered infrastructure built to solve academic and real-time operations.

Primary Project

UE Connect

University Social Network & AI-Powered Academic Platform

A comprehensive university social network and academic mentoring platform integrating event-driven communication feeds with a self-hosted semantic advising AI pipeline.

Problem

University environments often lack integrated systems for students to communicate in real time, collaborate, and receive automated academic advising, leading to information fragmentation and advisor burnout.

Solution

UE Connect merges a high-fidelity community feed, direct websocket messaging, and advisor mentoring structures with a self-hosted semantic AI retrieval subsystem in a single platform.

Core Systems
Social Feed & Interaction

Event-driven posts, nested replies, media uploads, and custom user tag mentions (@mentions).

Real-time Communication

Low-latency direct messaging, visual presence trackers, and instant push notifications using Laravel Reverb and Redis brokers.

Academic Mentoring

Direct advisement queues matching student cohorts with designated faculty advisors for structured mentorship.

AI & Academic Intelligence

Integrates the HCMUE RAG Assistant, a self-hosted Retrieval-Augmented Generation pipeline. It parses institutional documents, indexes vector chunks into Qdrant using BGE-M3 embeddings, and uses Gemini to answer policy queries with strict citation references.

HCMUE RAG AssistantQdrantGemini IntegrationRAG PipelinePDF ProcessingCohort DetectionCurriculum AnalysisAcademic Search
Technical Impact

Successfully migrated institutional curriculum details into a semantic vector index, resolving student query delays and automating academic advisement workflows.

1,290+Academic PDFs
11M+Tokens Indexed
15K+Vector Chunks
Technology Stack
LaravelPHPMySQLPythonGemini APIQdrantBGE-M3RAG PipelinesRedisLaravel ReverbWebSocketsReactTypeScriptTailwind CSSLivewireAlpineJS

Supporting Systems

Independent modules powering specific operations within the platform.

Realtime Chat & Notification

A standalone backend service handling client WebSocket channels, message persistence, presence tracking, and server event broadcasting.

Solution

Configured a low-latency socket layer utilizing Laravel Reverb alongside Redis pub/sub to handle websocket event distribution and queue caching dynamically.

Impact

Provides reliable real-time messaging, instant push notifications, and presence trackers with sub-second message delivery overhead.

LaravelRedisLaravel ReverbWebSockets

Community & Mention System

An interactive posting and communication package replicating standard high-fidelity social features like comments, nested replies, media uploads, and mentions.

Solution

Created custom regex-based text parsing filters that convert mentions to links, trigger asynchronous job workers for email/push notifications, and store uploads using standardized moderation checks.

Impact

Delivers Facebook-style posting features, clean nested replies, and immediate notifications when a user is tagged in posts or comments.

LaravelAlpineJSMySQLTailwindCSS

Engineering Capabilities

What I build, what problems I solve, and where I have hands-on experience.

AI & RAG Engineering

Built retrieval-based AI systems capable of searching, indexing, and reasoning over large academic datasets.

Highlights

  • Processed 1,290+ academic PDFs into a structured knowledge base.
  • Indexed over 11M tokens across 15K+ vector chunks in Qdrant.
  • Built end-to-end RAG pipelines with semantic search and citation-grounded answers.
  • Integrated Gemini for LLM orchestration and structured response generation.

Technologies

PythonGeminiQdrantBGE-M3EmbeddingsVector Search

Technical Expertise

A structured breakdown of core paradigms and the technologies I utilize to build retrieval-augmented services and web architectures.

Primary Capabilities

Core Expertise

Laravel Ecosystem

Designing service-layer MVC applications, database relations, task scheduling, and background queues.

AI & Retrieval

Engineering semantic retrieval systems, dynamic contextual chunking, vector database optimization, and LLM agent integration.

Realtime Systems

Orchestrating event broadcasting pipelines, state syncing, and high-frequency WebSocket messaging using Redis and Laravel Reverb.

AI & Retrieval

Vector database indexing and LLM pipelines driving semantic academic intelligence.

Gemini

LLM orchestration and structured semantic generation.

RAG

Retrieval-Augmented Generation pipeline structures.

Qdrant

Vector database powering academic search workflows.

Embeddings

Dense mathematical context representations of text.

Vector Search

Cosine similarity scoring and proximity database search.

BGE-M3

Multilingual embedding model mapping curriculum context.

Backend & Realtime

Service foundations and event-driven architectures powering systems logic.

Laravel

Primary framework for large-scale web applications.

Redis

High-performance cache, pub/sub messaging, and queue broker.

Laravel Reverb

WebSocket server broadcasting instant client events.

PostgreSQL & MySQL

Structured relational database management and transactional storage.

Frontend & Languages

Reactive UI rendering and primary software engineering languages.

Python

AI orchestration scripts, embedding generation, and data processing.

TypeScript

Typesafe frontend state management and reactive components.

React & Next.js

Frameworks for building interactive user experiences.

Livewire & AlpineJS

Lightweight reactive frontend binding directly to backend states.

Let's Build Something Meaningful.

Open to software engineering, AI engineering, open-source, and product-building opportunities. Whether you want to collaborate on a system or just discuss retrieval-augmented generation pipelines, feel free to reach out.

Available For
  • Backend Engineering
  • AI & RAG Systems
  • Full-Stack Development
  • Open Source Collaboration
GitHubLinkedIn
© 2026 Nguyễn Tuấn Khôi. All rights reserved.vietnam // callm3tk