Nº 01 · Portfolio proof sheet · 1 of 1

Nº 01 · Hero, display proofSheet 1/1 · EN

Igor Albanezi Conceição

Senior Frontend Engineer, currently an Applied AI Engineer on an AI-native team.

Nine years building interfaces, from BMW assembly-line software to AI-native product teams. This page is the proof of claim: set, measured, and annotated like a specimen of its own design system.

10-second read

  • 9 years frontend — Angular, TypeScript, RxJS, design systems
  • Apr 2026: internal move to Applied AI Engineer — agents + evals
  • Porto, PT · open to hiring conversations

02 · Projects, case studies

Projects

P.01

Assembly-line software, BMW group

Features, libraries and mentoring on BMW assembly lines.

Disclosure · attributable — Employer and domain named; work described at feature scope, no client material shown.

Story

Shipped features on software used on the assembly line itself — production-critical by default

Built shared libraries and component foundations adopted by teams beyond my own

Mentored frontend developers — reviews, pairing, onboarding

Contributed to architecture and design discussions

Angular · TypeScript · RxJS

P.02

Digital-banking platform, Inter

Front-end delivery for regulated banking products.

Disclosure · attributable — Employer and domain named; no client material shown.

Story

Built customer-facing features inside a large, regulated codebase

Shipped within cross-functional product teams under banking-compliance constraints

P.03

Consulting delivery, Aubay Portugal

Embedded front-end delivery inside client teams.

Disclosure · client confidential — Employer named, client and product abstracted.

Story

Front-end delivery embedded in client teams across industries

Became productive in unfamiliar codebases fast — consulting’s actual job

P.04

This specimen — the site itself

Ships with Lighthouse 100 ×5 and LCP under 1.5s, enforced by CI: this page’s own engineering is the proof of claim.

Disclosure · attributable — Code walkthrough available on request.

Context

For anyone who wants to know how I work. After reading it, they should trust that I ship carefully, keep learning, and can work with AI agents as part of how I build.

Numbers

Lighthouse categories at minimum score 1
100 ×5
Source · Lighthouse CI
Largest Contentful Paint
< 1.5s
Source · Lighthouse CI
Read the case study: This specimen — the site itselfClose the case study: This specimen — the site itself2 problems · 3 decisions · story · lesson

Problems

  1. I needed to design, review and ship the site from anywhere, without sitting at an IDE.
  2. I had to keep the code quality high while agents wrote most of the code.

Decisions

  1. D.01

    Build with cloud agents instead of an IDE

    It frees my time for architecture, deliverables and the product.

    Trade-off I'm further from the code.

  2. D.02

    Static export with Sanity content fetched at build time

    The page ships as pre-rendered HTML on Cloudflare Pages, with no server runtime to keep fast or secure.

    Trade-off A content edit needs a rebuild before it goes live.

  3. D.03

    Performance and accessibility targets enforced in CI

    Lighthouse CI assertions fail the build, so the numbers on this sheet are not aspirational copy.

Story

I started with OpenCode in the terminal, moved to Codex on Cloud, and now run a team of Grok Bots for UX, implementation and architecture review. Nothing merges until the CI gates, a thermo-nuclear review and all three bots pass it.

Lesson

Optimizing from day one felt disciplined, but it was premature. Make it work first, then make it fast.

Next.js · TypeScript · Sanity · Cloudflare


03 · The shift, transition record

The shift

In April 2026 I moved inside Critical TechWorks — same employer, a different mandate — from senior frontend to an R&D project on the AI-native team. The focus: orchestrating agents.

The frontend background is the method, not the past. Nine years of making complex systems legible in a browser is exactly what an AI product needs at the interface layer: model output has to be rendered, steered, and judged — and evals are just tests with less deterministic subjects.


04 · About, revision history

About

Career revisions, reverse chronological
RevisionYearsRoleOrganizationNote
R.6Apr 2026 – presentApplied AI EngineerCritical TechWorksInternal move; AI-native R&D — agent orchestration
R.5Sep 2021 – Apr 2026Senior Frontend DeveloperCritical TechWorksBMW assembly-line software: features, libraries, architecture
R.42021 – 2022Front-End DeveloperAubay PortugalPorto
R.32021Front-End DeveloperInter
R.22020 – 2021Application DeveloperCESARRecife
R.12017 – 2020Frontend → Advanced Frontend DeveloperAccentureRecife

Nine years, one thread: making complex systems legible in the browser. Assembly lines, banks, consultancies, and now AI-native teams, where the interface has to earn trust in what a model says.


05 · Skills, character set

Skills

Principal

daily, production-proven

  • Angular
  • TypeScript
  • RxJS
  • HTML/CSS
  • Accessibility
  • Design systems

Working

shipped with, comfortably

  • React
  • Next.js
  • Node
  • Testing
  • CI/CD
  • Performance

Currently proving

the AI-native practice

  • LLM APIs
  • Agentic tooling
  • Evals
  • Streaming UI
  • Prompt architecture

06 · Contact, colophon

Contact

Hiring conversations welcome; happy to walk through the code behind this page.

06.1 · Contact, send a message