Let's talk

Open to jobs, freelance projects, custom AI builds & co-founding

Hi, I'm Habi.I build AI products

AI EngineerBackend EngineerFull Stack Developer

4+ years turning ideas into production software, about 3 of them in production AI, with Python, FastAPI, Django and React / Next.js — from APIs and data pipelines to AI agents and RAG systems. Hire me, bring me a project, or let's build a product together.

Ways to work with me See what I've built
Work mode
Remote worldwide · Hybrid / on-site in Indonesia
Open to relocate
NZ · AU · SG · MY · TW · JP · US
Languages
English · Indonesian

01Ways we can work together

Whether you're hiring, have a project, need an AI tool built for your business, or want a technical partner.

For companies

Hire me

Full-time or contract roles as an AI Engineer, Backend Engineer or Full Stack Developer. Remote for teams anywhere; hybrid / on-site in Indonesia, or relocating abroad.

Discuss a role →
For project owners

Build your project

Freelance delivery of backend APIs, web apps and AI features, from requirements and a working prototype to deployment and handover.

Tell me about your project →
For businesses

Custom AI solutions

An AI tool built around your data and workflow, such as:

  • Market & competitor research agents
  • Lead research & data enrichment
  • Chatbots over your documents (RAG)
  • Automated reports & documents
  • Web & PDF data extraction
  • Transcription & workflow automation
Request an AI solution →
For founders & builders

Build something together

Looking for a technical partner or co-founder for an AI product? I'm open to co-building products, startups and community or open-source projects.

Let's build together →

How I run a project

  1. 1DiscoverUnderstand the problem, users and constraints
  2. 2PrototypeA quick proof of concept to validate the approach
  3. 3BuildProduction-grade APIs, UI and AI pipelines
  4. 4Ship & measureDeploy, monitor, tune cost and quality
  5. 5Hand overDocs and a clean handover, or continued support

02Projects & case studies

Product descriptions use publicly available information. Diagrams are simplified, generic views, not internal designs. Source code, prompts, data and client information stay confidential.

Featured

Growise · 2024 – present · AI + Backend

TAM AI

AI go-to-market platform for B2B export sales teams: describe a target market in plain language and AI agents research companies, decision-makers and buying signals across multiple data sources.

  • Agent architecture: multi-step reasoning, tool calling, structured outputs
  • Task management and infrastructure decisions
Agentic AIPythonFastAPILLMs
Product site ↗

Growise · 2023 – present · AI + Backend

Multi-LLM RAG Modules

Containerized, reusable Retrieval-Augmented Generation modules with pluggable LLMs and embedding models, shared across product domains.

  • Isolated per-user indexes to prevent cross-user data leakage
  • Re-engineered retrieval: ~30% lower latency and ~60% higher reasoning accuracy vs. earlier versions
RAGRerankingVoyageAIPineconeMongoDB

ITS · 2022 – 2023 · Full Stack + AI

3DVT: Ultrasound Vein Segmentation

Healthcare AI research prototype: a web app that segments deep veins from ultrasound images to support Deep Vein Thrombosis diagnosis, with interactive 3D visualization.

  • Built the React + Django app and backend serving the research team's U-Net (TensorFlow) model
  • Model preloader for low-latency inference; JWT auth hardening
ReactDjangoTensorFlowThree.js

More projects

03Experience

  1. Nov 2023 – Present

    AI Engineer

    Airconcur (Growise) · Taipei, Taiwan · Remote

    Joined as a freelance AI Module Developer; hired full-time in May 2024.

    • Architect and ship agentic and RAG-based AI microservices in Python (FastAPI) with multiple LLMs and embedding models, working directly with C-level leaders.
    • Up to 70% lower cost and 4× faster than a baseline tool-calling agent, through semantic, prompt and tool-output caching, data canonicalization, and context & memory management.
    • Re-engineered the retrieval layer (Voyage AI embeddings, reranking, data cleanup and de-duplication): ~30% lower latency and ~60% higher reasoning accuracy vs. earlier versions.
    • Built containerized, multi-LLM RAG modules with isolated per-user indexes to prevent cross-user data leakage.
    • Added AI safety layers: PII scrubbing, prompt-injection prevention and moderation middleware.
    • Delivered AI services for enterprise customers, including Advantech; 2× faster builds after reworking CI/CD.
  2. May 2023 – May 2024

    Full Stack Developer / Software Engineer

    PT Mandiri International Technology · South Jakarta

    • Owned the full project lifecycle, from requirement gathering to deployment.
    • Built customer-facing web apps (React, Next.js) and a cross-platform mobile app on a Django REST backend.
    • IoT device–backend communication over HTTP and MQTT; tuned ORM and queries for 100+ concurrent requests.
  3. Aug 2022 – Mar 2023

    Full Stack & Backend Developer (Healthcare AI)

    Institut Teknologi Sepuluh Nopember (ITS) · Freelance · Surabaya

    • Built the React–Django app and backend serving the research team's AI model for ultrasound vein segmentation and 3D visualization.
    • Replaced legacy auth with JWT and refactored API viewsets for performance and reliability.
  4. Apr 2022 – Apr 2023

    Full Stack Web Developer

    LnData Inc. · Surabaya

    • React / Django products, secure APIs and Airflow data pipelines for social-media platforms.
    • Managed deployment servers, monitored system load and resolved security issues.
  5. 2017 – 2021

    B.Eng. Computer Engineering

    Institut Teknologi Sepuluh Nopember (ITS) · Surabaya

04Skills & stack

AI Engineering

LLM features that survive production: agents with tool calling, RAG over your own data, and the evaluation and cost tuning that keep them reliable.

  • Agentic workflows & multi-agent orchestration
  • RAG: embeddings, vector DBs, reranking
  • Token, latency & model-cost optimization
OpenAIAnthropicLangChainLlamaIndexPinecone
Backend Engineering

Python services and APIs that are secure, observable and fast under load — the part of the stack I've spent most of my career in.

  • REST APIs & microservices (FastAPI, Django)
  • Async jobs, queues, caching, data pipelines
  • Query tuning, auth (JWT) & CI/CD
PythonFastAPIDjangoPostgreSQLMongoDBRedisDocker
Full Stack Development

End-to-end delivery, from requirement gathering to deployment: responsive web apps on a solid backend, plus a cross-platform mobile app shipped in production.

  • React / Next.js frontends with TypeScript
  • Redux Toolkit, RTK Query, Tailwind, MUI
  • Owning features across frontend, API & DB
ReactNext.jsTypeScriptTailwind CSSDjango REST

Also worked with

Pydantic AIGoogle GenAIVoyageAIChromaDBCeleryRabbitMQApache AirflowMinIOAzureDjango ChannelsWebSocketsMQTTPlaywrightMaterial UIThree.jsTensorFlowC++ESP32 / Arduino

Open to new stacks

I've moved between Django, FastAPI, React, IoT hardware and LLM tooling as projects needed. I'm happy to pick up whatever your team runs on, such as Node.js, Go, AWS, GCP or Kubernetes.

05Let's talk

A role, a project, a custom AI tool or a product idea: send me a short note about what you have in mind and I'll get back to you.

Or start with a topic:

Job opportunity Project inquiry Custom AI solution Collaboration
Download CV (PDF) LinkedIn GitHub

Growise (Airconcur) · AI Engineer · 2024 – present

TAM AI — AI go-to-market platform

The product

B2B export sales teams spend hours finding which overseas companies buy their product, who the decision-maker is, and whether an email actually works. TAM AI lets a user describe a target market in plain language; AI agents then research it across a global company database, US customs trade records (by HS code), LinkedIn decision-maker data and government tenders, verify email deliverability, and return a structured, exportable list that can be monitored for new buying signals.

How it works (simplified, public view)

Plain-language market brief→AI agent plans the research→Queries data sources & tools→Cross-checks & structures results→Verified list + signal monitoring

My role

Engineering focus across my Growise role

PythonFastAPILLM agentsTool callingStructured outputsAsync
Public product site ↗

Growise (Airconcur) · AI Engineer · 2024 – present

AI Workflow — AI automation for business teams

The product

Growise Workflow is an AI automation tool that sequences tasks and generates content, so business teams can turn repetitive, multi-step work into a reusable workflow instead of prompting a chatbot by hand every time.

How it works (simplified, generic view)

Goal / trigger→Break into steps→Each step calls an LLM or tool→Validate structured output→Result / next step

My role

PythonFastAPIMulti-LLMOrchestrationStructured outputs

Growise (Airconcur) · AI Engineer · 2024 – present

Scraper AI — web data for people and agents

The product

Growise's Scraper lets non-technical users extract data from websites without writing code. The same need shows up inside AI agents: they need clean, structured content from web pages and documents before they can reason over it.

How it works (simplified, generic view)

URL / site→Crawl pages (browser automation)→Parse HTML & PDFs→Clean, LLM-ready text / fields→Agent or export

My role

PythonPlaywrightWeb scrapingPDF parsingData pipelines

Growise (Airconcur) · AI Module Developer → AI Engineer · 2023 – present

Multi-LLM RAG Modules

The problem

Several AI products needed to answer questions over customer-specific data, with LLMs and embedding models tailored per client, and without ever mixing data between users.

How it works (standard RAG pattern)

Documents→Chunk & embed→Per-user vector index→Retrieve & rerank→LLM answer with context

What I built

Results

PythonLlamaIndexLangChainVoyageAIRerankingPineconeChromaDBMongoDBDocker

Growise (Airconcur) · 2023 – present

Applied AI Services

What they do

A set of microservice-based AI services, tailored per client, for enterprise customers:

What I did

Results

PythonFastAPIOpenAIAnthropicGoogle GenAIVoyageAI

PT Mandiri International Technology · Full Stack Developer · 2023 – 2024

IoT Platform & Site Monitoring

The work

I owned projects end to end, from requirement gathering with the project director to deployment: customer-facing websites, a cross-platform mobile app, and the backend that IoT devices and a site-monitoring system talk to.

How it works (simplified)

Devices (ESP32, Arduino, Orange Pi)→HTTP & MQTT→Django REST + Celery workers→PostgreSQL · MongoDB · Redis · RabbitMQ→Web & mobile apps

Results

DjangoDRFCeleryReactNext.jsRTK QueryMQTTC++

Institut Teknologi Sepuluh Nopember (ITS) · Freelance · 2022 – 2023

3DVT — Ultrasound Vein Segmentation

Research prototype, not a medical device.

The problem

The research team needed a web app to support Deep Vein Thrombosis (DVT) diagnosis: segment veins from ultrasound images automatically and show them in 3D, as part of a blood-clot volume determination system.

How it works (simplified)

Ultrasound images→Django API→U-Net segmentation (TensorFlow)→3D vein model (GLTF)→React + Three.js viewer

What I built

ReactDjangoTensorFlowU-NetThree.jsJWT

LnData Inc. · Full Stack Web Developer · 2022 – 2023

Social Media Data Platforms

The work

My first role as a software engineer: frontend and backend development for social-media data products, working with multinational teams.

Results

ReactDjangoCeleryApache AirflowRedisSQL

Undergraduate thesis · ITS Surabaya · Published in Jurnal Teknik ITS, 2021

Handwashing Motion Classification with CNN

The problem

During the COVID-19 pandemic, many people still washed their hands too quickly or skipped steps, and nobody can watch public sinks 24/7. The goal: use a camera and deep learning to recognize each handwashing step and check whether it was done properly.

How it works

Handwashing video→Frame extraction→Data augmentation→EfficientNet-B0 classifier→Moving-average smoothing→Labeled output video

Demo video

Left: raw input. Right: the predicted handwashing step written on each frame. Watch on YouTube ↗

What I did

Results

PythonTensorFlowEfficientNet-B0Computer visionData augmentation

Habibul Rahman Qalbi, Eko Mulyanto Yuniarno, Reza Fuad Rachmadi. Klasifikasi Gerakan Cuci Tangan Berbasis Convolutional Neural Network (CNN). Jurnal Teknik ITS, 10(2), 2021.

Read on Neliti ↗DOI ↗