Part 5 ยท 25 minutes

Make the relevant evidence easy to find.

A profile or resume does not need to contain everything. It must help the reader quickly understand the target role, strongest evidence, scope of work and credible outcomes.

Ten-second scan

If a recruiter reads only your headline, top third of the resume and first two experience bullets, what role will they believe you are suited for?

LinkedIn profile audit

Headline

Target role + specialization + evidence. Avoid an unstructured list of every tool learned.

About

Present focus, two or three proof points, and the direction being pursued.

Experience

Problems, ownership, decisions and outcomes rather than copied job descriptions.

Featured

Two to four strong artifacts: project, talk, paper, demo or repository.

Skills

Relevant skills supported by actual work or projects.

Settings

Accurate titles, locations, work modes, contact details and job preferences.

Resume principles

Weak bullet

Worked on a machine learning model for recommendations using Python.

The problem, ownership, method, scale and result are all unclear.

Stronger bullet

Improved top-10 recommendation recall by 11% by adding session-level features and time-aware negative sampling across 4.2 million interactions.

Every number and claim must be truthful and defensible.

Useful pattern: Built or improved [system] using [method], achieving [measured result] on [scale], leading to [outcome].

This is a thinking template, not a sentence generator. Not every bullet needs every field, and a fabricated metric is worse than an honest qualitative result.

Resume defence sheet

Prepare this for every significant project or achievement. Anything written on the resume is permission for the interviewer to ask deeper questions.

Problem

What user or business decision mattered?

Ownership

What did you personally design, code or decide?

Data

How was it collected, labelled, split and checked?

Model

Why this baseline and model family? What alternatives were considered?

Metric

Why did the metric represent the real cost?

Result

How was the result measured and at what scale?

Failure

What underperformed or surprised you?

Production

How was the output deployed, monitored or consumed?

Reflection

What would you change with another month?

Formatting and templates

Previous: System designNext: Jobs and JDs