David Angel
Product engineer · David AngelBased in Lima, Peru

I turn complex
problems into
useful products.

I connect user discovery, interface design, and hands-on engineering. I lead the vision and build the critical features, with an eye for the details people actually experience.

Let’s build something that matters

Selected work

01 — 03
01 / PAPAYAProduct & engineering

From school data
to a tutoring plan.

A decision tool that connects complex accountability rules to a plan people can compare, question, and act on.

Papaya / Accountability
INTERACTIVE RECONSTRUCTION

One goal. Two ways forward.

PLANNING EXAMPLE

Compare the trade-off when a school’s enrollment changes.

Projected D2A score75 / 100Target: 75
Tutoring hours168 hResource reference: 132 h
Student–subject pairs12Proposed for tutoring

Baseline and conditional projections

Previous baseline61
No improvement59
Intervention75
The target stays. The resources change.

This example reaches the projected target with 36 additional tutoring hours.

The previous baseline may describe an earlier roster. “No improvement” assumes unchanged performance levels; projections are not guaranteed outcomes.

Synthetic data · fixed illustrative scenariosOriginal scenario preserved
Explore a reconstructed planning decision. These synthetic examples explain the experience; they are not measured school outcomes.FIG. 01
Product discovery · UI/UX · Optimization · Full-stackRead the accountability case
A new roster. A deliberate new decision.
ORIGINAL SCENARIOKeep the context.

Earlier plan preserved

REPLACEMENTReplan with intent.

Review, then present

Source → Compare → Present
02 / PAPAYAA closer look

“The data changed.”
So the workflow had to.

How an ambiguous request became a deliberate way to replace a roster, compare the consequences, and preserve the earlier plan.

Inside the product decision
angelstack / guard integrityILLUSTRATIVE FIXTURE
// A known failure, introduced on purpose
window.confirm("Delete item?");

expected  exit 1 · native dialog detected
clean     exit 0 · no banned controls
A check earns trust by catching the failure.
03 / ANGELSTACKOpen source

Product judgment,
made executable.

Skills and checks for building with coding agents. Interface consistency, accessibility, and known failure cases that make quality repeatable.

Explore angelstack
How I work

An engineer’s head.
A designer’s eye.

I started in product because I wanted to understand the problem. I moved into engineering because I wanted to build the answer.

01

Find the real problem.

Sit with the people doing the work. Question the request, understand the constraints, and uncover the decision the product needs to support.

02

Give the experience a clear shape.

Design the flow, the screens, and the system behind them. Make the important action obvious and the difficult decision understandable.

03

Build it. Stay close to it.

Plan with the team, personally build the critical path, and use working demos to keep the implementation true to the experience.

The next questions

Work in progress
Planned engineering thesis

How good is the recommendation?

I’m developing the accountability platform’s optimization problem into an operations-research thesis. The next step is to define how to evaluate the model and its trade-offs rigorously.

The research direction
Upcoming AI product work

AI that earns its place in the product.

I’m exploring retrieval, tool use, and evaluation for educational workflows. The aim is a useful, testable experience, with a clear distinction between what the model suggests and what the system knows.

A good product starts with a conversation

Have something
worth building?

Let’s talk about product engineering opportunities and the problems your team is taking on.

Email David Connect on LinkedIn