New QuantDesk is now in beta. Try it

An independent AI lab · Bogor, Indonesia

Small AI tools,
built with care.

RangkaLabs designs and ships focused AI products, and shows how each one is built and tested. One lab, a few experiments, and a flagship that's ready to try.

rangkalabs / projects
QuantDeskStock analysis Beta
ScholarTrackGrounded LLM extraction Experiment
RAG eval dashboardRetrieval + evaluation Idea
$ curl -I https://rangkalabs.cloud
HTTP/2 200 · served by Caddy
LAB NOTE · SOON ScholarTrack: no quote, no value
WHAT THE LAB WORKS ON Stock analysis Grounded LLM extraction Retrieval + evaluation

01 / Flagship

One product, done properly.

The first RangkaLabs product, and the one to try first.

Beta Stock analysis

QuantDesk

Stock analysis for everyday investors. Instead of one score, it shows where models that share no data agree, where they contradict each other, and what none of them can see.

  • Four independent lenses — order flow, price trend, intrinsic value and accounting quality, each read from different data and shown side by side.
  • A written verdict, not a score — says where the lenses agree and where they disagree, and what the app can't tell you about this company.
  • Scan and rank an index — ranks up to 250 stocks as percentiles, and shows its own backtest finding that the ranking doesn't predict returns.

An analysis tool, not investment advice.

QuantDesk analysing NVDA: Flow reads Quiet, Trend reads Uptrend, Value reads Above model range and Quality reads Concerns, for an overall verdict of Mixed signals.

02 / Approach

How the lab works.

Four rules every RangkaLabs project follows, from the first experiment to launch.

One problem per tool

Each product does one job well instead of ten jobs badly.

Measured, not guessed

AI features ship with evaluations, so quality is a number, not a feeling.

Built in the open

Every project gets a lab note: what worked, what didn't, and why.

Honest about limits

Clear labels for what's live, what's in beta, and what's still an experiment.

03 / The lab

Experiments in progress.

Some will become products. Some won't. Each one gets a write-up either way.

Idea

RAG with an eval dashboard

Answers grounded in source documents, with a public dashboard that measures how often they're right.

retrieval evals dashboard
Experiment

ScholarTrack

An LLM reads scholarship funders' own websites and pulls out deadlines, coverage and eligibility. A value is kept only if its verbatim quote is found on the page.

Try it — ScholarTrack

04 / How projects grow

From idea to product.

Every project carries one of these labels, so you always know what you're looking at.

  1. 1
    Idea

    A problem worth exploring. Nothing built yet.

  2. 2
    Experiment

    A working prototype, tested and measured. May be dropped.

  3. 3
    Beta

    Real people use it. Rough edges are expected and fixed fast.

  4. 4
    Live

    Stable, supported, and ready to rely on.

05 / Roadmap

What's happening now.

Plans change. This is the current best guess.

Now
  • QuantDesk beta
  • ScholarTrack experiment
Next
  • First lab notes published
  • QuantDesk public launch
  • RAG eval dashboard

06 / Lab notes

How things get built.

Short write-ups on architecture, evaluation, and what didn't work.

  • COMING SOON ScholarTrack: no quote, no value grounding
  • COMING SOON Building QuantDesk: the architecture architecture
  • COMING SOON Setting up a one-person lab on a small VPS infra
Farhan Rangkuti, arms crossed, on a rocky beach

07 / About

Built by Farhan Rangkuti.

I'm the founder of RangkaLabs, building AI and ML products from Bogor, Indonesia. By day I'm a bioinformatics technician at PT SMART Tbk, building genotype data pipelines and self-hosted AI tools for oil palm research. I studied Informatics at Telkom University, where my thesis on graph neural networks won Best Paper at ICICyTA 2024.

Rangka means “frame” in Indonesian: the structure everything else is built on.

08 / FAQ

Questions, answered.

Something else? Ask by email.

What is RangkaLabs?

An independent AI lab run by Farhan Rangkuti in Bogor, Indonesia. It builds small AI products and publishes how they're made.

Can I try the stock app?

It's in beta. Try QuantDesk.

Is the stock app investment advice?

No. It's an analysis tool to help you research. Investment decisions stay yours.

Do you take on projects or collaborations?

Yes. Email me at farhanrkt@rangkalabs.cloud with what you have in mind.

How do I follow along?

Lab notes are posted here. Email to be told when new ones are out.

09 / Contact

Have a problem worth building for?

Email is the fastest way to reach me.

farhanrkt@rangkalabs.cloud