// Apriani Nur Raina · Full Stack AI Engineer · Jakarta, Indonesia
Full Stack AI Engineer in Jakarta building Retrieval-Augmented Generation (RAG) pipelines, conversational voice AI and computer vision systems — Python, FastAPI and LangGraph underneath, React and TypeScript on top. Four AI products shipped, and 75% cut from inference infrastructure cost.
Jakarta, Indonesia·open to Full Stack / AI Engineer roles
const impact = {
// aiProductsShipped
Voice AI, computer vision and recommendation systems in production
// yearsEngineering
Two years frontend, then full stack AI engineering
// inferenceCostCut
Saved by re-architecting the AI inference pipeline
// pageLoadImproved
Faster loads through frontend and API optimisation
};
// about.md
I build production-grade AI applications end to end — from React interfaces down to FastAPI services and the LLM orchestration underneath. My work covers Retrieval-Augmented Generation (RAG) pipelines, conversational voice AI, computer vision and recommendation systems, and the part I care about most is the unglamorous half: retrieval quality, latency, cost and reliability once real users arrive.
Before specialising in AI I spent two years as a frontend developer building React and React Native interfaces at scale. That background is why I own the whole path — an AI feature has to feel fast and trustworthy to use, not just score well offline.
// projects.json
Four production AI systems across voice, vision and recommendations, plus the automation that triages this site's own enquiries. Each write-up states the problem, the approach, the stack, and what actually changed for the people using it.
$ n8n execute --file contact-triage.json
Every enquiry that reaches me gets validated, classified and routed before it ever hits my inbox — recruiter, freelance, collaboration or spam, each with a priority score and a destination. Built as an n8n workflow; try it below.
Pick a preset or write your own, then run the workflow.
// The workflow itself is self-hosted n8n. This demo posts to a serverless endpoint running the same rule set, so it stays live without an always-on instance — point N8N_WEBHOOK_URL at a running n8n and the identical request executes there instead.
// skills.yaml
Split the way I'd describe it in an interview — what I ship with daily, what I'm comfortable working with, and what I'm deliberately learning next. Nothing is padded.
llm_and_rag:
computer_vision:
languages:
backend:
frontend:
data:
orchestration:
data_infra:
deployment:
llmops:
agents:
cloud_ai:
The credentials I'm working toward, in the order I plan to take them. Listed as a plan, not as qualifications already held.
see the full breakdownAWS Certified AI Practitioner
Amazon Web Services
AWS Certified Machine Learning Engineer — Associate
Amazon Web Services
Professional Machine Learning Engineer
Google Cloud
Azure AI Engineer Associate (AI-102)
Microsoft
$ git log --stat
// mar 2025 — present
// dec 2024 — present
// sep 2023 — feb 2025
$ ls ~/awards
Five national and international UI/UX design placings between 2022 and 2023, including two first places. Each card links to the organiser's own announcement.
IHRD Tharang · India
// gold
Feb 2023
SE-Research Innovation Challenge
// gold
Sep 2022
IT Festival IPB 2023
// silver
Sep 2023ESD Laboratory Challenge
// bronze
2022
I/O Festival UNTAR 2023
// finalist
202301 / 05
// contact.sh
I'm open to AI Engineer, Full Stack AI and Machine Learning Engineer roles, and take on freelance AI work. Tell me what you're building and I'll tell you honestly whether I'm the right fit.
#!/bin/bash