Career Document System Build Plan

Source: career-document-system-build-plan.md (ingested 2026-03-28)

Phase 3: Career Document System — Build Plan

DIAGNOSTIC FIRST: What We're Actually Solving

Before building anything, let me name the real problem clearly.

You are not a standard candidate. You sit at a rare intersection — computational physics, data science, research, and likely broader technical + analytical depth. The failure mode for someone like you is one of three things:

Failure Mode 1: The Over-Specialist Profile reads as "physics PhD who codes" and gets filtered out of industry roles by recruiters who don't know what a physicist does.

Failure Mode 2: The Chameleon Without Spine You sand off everything distinctive trying to fit every job description and become unrecognizable. You look like everyone else. You get treated like everyone else.

Failure Mode 3: The AI-Obvious Profile Every sentence is perfectly structured, uses the exact same verb list (spearheaded, leveraged, architected), and has the cadence of a language model. Recruiters increasingly recognize this and it destroys trust before you get a call.

The goal is to build a system that avoids all three — a profile that is genuinely you, strategically positioned, and platform-appropriate without being platform-captured.

ARCHITECTURE: The Career Document System

career-platform/
├── core-identity/
│   ├── master-narrative.md          # Who you are in your own words — source of truth
│   ├── skills-inventory.yaml        # Every real skill, rated honestly, categorized
│   ├── experience-bank.yaml         # Every role/project — raw, unformatted
│   ├── achievement-bank.yaml        # Specific results, numbers, outcomes
│   └── voice-profile.md             # How you actually write/speak — anti-AI anchor
│
├── documents/
│   ├── resumes/
│   │   ├── base-resume.md           # Full master resume (longer than any submission)
│   │   ├── research-track.md        # Academia/national lab/R&D variant
│   │   ├── industry-ds-track.md     # Data science / ML engineering variant
│   │   ├── quant-track.md           # Finance/quant/trading variant
│   │   └── generalist-track.md      # Broad technical roles, startups
│   ├── cvs/
│   │   ├── academic-cv.md           # Full CV for academic/postdoc applications
│   │   └── research-cv.md           # Trimmed CV for industry research positions
│   └── cover-letters/
│       ├── templates/
│       │   ├── research-template.md
│       │   ├── industry-template.md
│       │   └── cold-outreach-template.md
│       └── sent/                    # Archive of submitted letters
│
├── platforms/
│   ├── linkedin-strategy.md         # Profile sections, content strategy, connection approach
│   ├── platform-matrix.yaml         # Which platform for which role type
│   └── profiles/
│       ├── linkedin.md              # Full LinkedIn content (copyable)
│       ├── handshake.md
│       ├── indeed.md
│       ├── dice.md
│       └── glassdoor.md
│
├── role-targeting/
│   ├── role-registry.yaml           # Tracked roles/companies of interest
│   ├── fit-matrix.md                # Which resume variant → which role type
│   └── application-log.yaml         # Every application, status, follow-up
│
└── refinement/
    ├── feedback-log.md              # What worked, what didn't, interview feedback
    ├── version-history/
    └── checklist.md                 # Pre-submission quality gate

PHASE 3A: FOUNDATION — The Core Identity Layer

This is the most important step and most people skip it entirely. Before we write a single bullet point, we build the source of truth.

Step 1: Master Narrative

This is not a summary for employers. It is your internal document that answers:

What problems do you actually like solving?
What is the through-line of your work across projects and roles?
What makes your background genuinely unusual — not just "rare" in a buzzword sense?
What do you want your next role to feel like, and what are you moving toward?
What do you never want to do again?

This becomes the filter for everything else. Every resume variant, every platform profile, every cover letter gets checked against this.

Step 2: Skills Inventory (Honest, Not Aspirational)

# skills-inventory.yaml schema
skill_id: string
category: [programming | statistical | domain | tools | soft | research-method]
name: string
honest_level: [learning | functional | proficient | strong | expert]
evidence: [list of specific projects/outcomes that prove this level]
years_active: number
last_used: date
notes: string   # anything that complicates the simple label

The honest_level field matters. Do not write Expert for things that are Proficient. Interviewers will find the boundary, and calibrated self-assessment is itself a professional signal.

Step 3: Achievement Bank

Every bullet point in every resume must be sourced from this bank. The format is:

What you did + How you did it + What happened as a result (with numbers where real)

Not: Developed machine learning models for data analysis Yes: Built ensemble classifier combining gradient boosting with physics-informed features, reducing false positive rate by 34% on [domain] dataset — deployed in [context]

Not: Conducted research on [topic] Yes: Derived analytical solution for [specific problem], reducing simulation runtime from 6 hours to 23 minutes by replacing numerical integration with closed-form approximation

The numbers do not have to be dramatic. They have to be real and specific. Specificity is the anti-AI signal.

PHASE 3B: DOCUMENT VARIANTS

The Resume vs. CV Decision Tree

Applying to:
  Academic position (faculty, postdoc) → Full Academic CV
  National lab research position       → Research CV (trimmed, but publication-complete)
  Industry R&D (Google Brain, etc.)    → Research-track Resume (2 pages max)
  Data science / ML engineering        → Industry DS Resume (1-2 pages)
  Quant / finance technical            → Quant Resume (1 page, metrics-dense)
  Startup / generalist technical       → Generalist Resume (1 page, outcome-focused)

Resume Variant Differentiation

Each variant is not a different document. It is a different lens on the same person.

Research Track Leads with intellectual contribution. Methods and theory front and center. Publications, conference presentations, and research outcomes given full weight. The audience knows what a physicist does.

Industry DS Track Leads with impact and scale. Methods mentioned by name (XGBoost, PyTorch, etc.) because recruiters run keyword searches. Computation and modeling translated into business or system outcomes. The audience may not know what a physicist does — show them what you built.

Quant Track Hyper-compressed. Numbers only. Statistical and mathematical sophistication shown through what you modeled and what it predicted, not through description of process. Anyone reading this already knows the tools.

Generalist Track Broadest. Shows you can communicate across domains, work in ambiguous problem spaces, move fast. Best for startups or roles at the intersection of technical and strategic work.

PHASE 3C: THE HYBRID IDENTITY SYSTEM

This is the core of what you asked about — not looking like a generic AI-output profile, not pigeonholing yourself as purely computational, while being strategically legible.

The Unique Profile Framework

Your identity across platforms should communicate one coherent thing that has multiple facets, not multiple competing identities.

The framework is: Anchor + Bridge + Signal

Anchor: The one thing that is undeniably, distinctively you. Your deepest domain. For you this is likely something in the physics/computation space — but stated in terms of what problems you can solve, not what degree you hold.

Bridge: The translation layer that connects your anchor to multiple role types. This is where your hybrid value lives. What does someone with your background bring to a data science team that a pure CS person cannot? What do you bring to a research team that a pure theorist cannot?

Signal: The evidence that you are a real person with genuine perspective, not a resume-generating machine. This is voice, specific projects with real texture, interests that inform your work, opinions about your field.

Anti-AI Voice Guidelines

This is not about writing worse. It is about writing like yourself.

ANTI-AI SIGNALS (real, not performative):
  ✓ Name the specific thing you found interesting about a problem
  ✓ Describe a failure or pivot and what you learned (resumes usually hide this — LinkedIn can show it)
  ✓ Use your actual vocabulary, not resume vocabulary
  ✓ Have one sentence that no one else could write about your work
  ✗ Never: "passionate about leveraging cutting-edge solutions"
  ✗ Never: "spearheaded cross-functional initiatives"
  ✗ Never: three consecutive bullets with identical grammatical structure
  ✗ Never: a summary that could describe 40,000 other people

PHASE 3D: PLATFORM STRATEGY

Different platforms serve different functions. Do not treat them as identical submission portals.

# platform-matrix.yaml structure

linkedin:
  primary_function: Passive discovery + network + inbound
  audience: Recruiters, hiring managers, collaborators, referrals
  profile_type: Narrative + selective depth
  update_frequency: Monthly minimum, active when searching
  key_differentiator: Your voice and perspective — not just credentials
  content_strategy: Occasional posts in your domain signal active expertise

handshake:
  primary_function: Campus recruiting + early-career pipelines
  audience: University recruiters, entry-to-mid roles
  profile_type: Clean, keyword-complete, GPA/research prominent
  note: If you are post-graduation, less critical but maintain it

indeed:
  primary_function: High-volume job discovery and application
  audience: Automated systems first, humans second
  profile_type: Keyword-optimized, ATS-safe formatting
  note: Less about standing out, more about not being filtered

dice:
  primary_function: Technical/engineering role discovery
  audience: Technical recruiters, especially contract/consulting
  profile_type: Skills-matrix emphasis, technical stack prominent
  note: Strong for data/ML/engineering adjacent roles

glassdoor:
  primary_function: Company research + passive discovery
  audience: Employers can see your profile if you apply
  profile_type: Keep consistent with LinkedIn, less detailed
  note: More valuable as research tool than application portal

arxiv/google-scholar:
  primary_function: Research credibility signal
  audience: Academic and industry research hiring
  note: Keep publications current — this is checked

LinkedIn Profile Architecture (Specific)

HEADLINE
  Formula: [What you do] + [for whom / in what domain] + [distinctive element]
  Not: "Physics PhD | Data Scientist | ML Researcher"
  Yes: "Computational physicist building predictive models at the boundary of [domain] and data science"
  
  The headline is the one line that shows everywhere. It must work for a recruiter 
  who knows nothing about physics AND a research director who knows everything.

ABOUT SECTION
  Paragraph 1: The anchor — what you actually work on, in plain language
  Paragraph 2: The bridge — what your background makes possible that is unusual
  Paragraph 3: What you are looking for / what problems excite you next
  Final line: One specific invitation — what someone should reach out about
  
  Length: 3-4 paragraphs. Not a wall of text. Not three bullet points.
  Voice: First person. How you would describe yourself to a smart person at a conference.

EXPERIENCE SECTION
  Each role: 3-5 bullets maximum
  Each bullet: sourced from achievement bank
  Format: ATS-safe (no tables, no columns, plain text structure)

FEATURED SECTION
  Use this. Most people ignore it.
  Put your best work here: paper, project, writeup, GitHub, anything with depth
  This is where the Anchor + Bridge + Signal system becomes visible

SKILLS SECTION
  Top 3 pinned skills should reflect your actual anchor, not generic terms
  Endorsements matter less than people think — keep it clean and accurate

RECOMMENDATIONS
  Even 1-2 strong, specific recommendations outweigh 50 endorsements
  Ask specifically: request that they describe a specific project or problem you solved together

IMPLEMENTATION SEQUENCE FOR PHASE 3

Step 1 (Do First, Before Any Document)

Build the core-identity layer. Specifically:

Start by filling out the master-narrative.md in your own words, no templates. I will help you refine it — but the raw material must come from you. This takes 30-60 minutes and makes everything else faster and more coherent.

Step 2

Build the skills-inventory and achievement bank from that narrative. We categorize, rate honestly, and add evidence.

Step 3

Write the base-resume together. Long version, no length constraints, everything documented. This becomes the source for all variants.

Step 4

Produce the academic CV from the base (additive — add publications, talks, coursework).

Step 5

Derive the role-specific resume variants (subtractive and reframing — not rewriting from scratch).

Step 6

Write LinkedIn About section using master narrative + bridge framing.

Step 7

Port simplified versions to platform-appropriate formats for each service.

Step 8

Build role-registry and fit-matrix — so when you find a position you like, there is a clear decision: which variant fits, what minor customization does the JD suggest, is there a cover letter needed.

QUALITY GATE: Acceptance Criteria

A resume variant is ready when:

[ ] Every bullet traces to a real entry in the achievement bank
[ ] No bullet uses generic resume vocabulary without specific evidence attached
[ ] The document could only describe you — not a category of person
[ ] ATS-safe formatting confirmed (no tables, headers, columns that break parsing)
[ ] The anchor is visible within 10 seconds of reading
[ ] The bridge is visible within 30 seconds
[ ] Length is appropriate to variant and audience

A platform profile is ready when:

[ ] Voice is consistent with master narrative
[ ] Headline works for both technical and non-technical audiences
[ ] Featured section or equivalent is populated with real work
[ ] No section is empty or left at platform default
[ ] Profile is consistent with resume variants without being identical

NEXT IMMEDIATE ACTION

To move forward, I need you to give me the raw material.

Specifically, answer these in whatever format feels natural — bullet points, stream of consciousness, however:

What is your actual academic and work background? Degrees, research areas, roles, projects — everything real.
What are the 3-5 projects or pieces of work you are most proud of, and why specifically?
What kinds of problems do you want to be working on in 2 years?
What domains or industries feel interesting to you beyond the obvious?
What do you not want on your profile — things that are true but do not represent where you are going?

Once I have that, we build the master narrative together, and every document follows from that foundation.