1  # Score Jobs Persona Template
  2  
  3  This template defines the candidate persona used by the scoring LLM to evaluate
  4  job fit. Fill each section following the instructions in the HTML comments. The
  5  filled output should be plain text saved to `prompts/score_jobs_persona.txt` —
  6  remove all HTML comments when done.
  7  
  8  Section headers with parentheticals (e.g., "DOWNGRADE SCORE, DO NOT AUTO-REJECT")
  9  stay in the filled output — the scoring LLM reads them as inline instructions.
 10  
 11  See `samples/` for filled examples (pick the closest domain match).
 12  
 13  **When filling:** Remove everything above the `---` line (this header) and all
 14  `<!-- ... -->` comment blocks. Only the section content goes into the output file.
 15  
 16  ---
 17  
 18  PROFESSIONAL IDENTITY
 19  <!--
 20  What goes here: 1-2 sentence elevator pitch — your domain, years of experience,
 21  and professional archetype. No PII (name, employer). Focus on what defines you,
 22  not a list of skills.
 23  Why it matters: The LLM uses this as the primary lens for evaluating job fit.
 24  Examples:
 25  - Tech: "AI/ML R&D Leader with 10+ years transforming NLP research into production systems. Applied Researcher & Builder."
 26  - Healthcare: "ICU Clinical Nurse Specialist with 12 years in critical care and trauma. Patient outcomes over process."
 27  - Marketing: "Growth Marketing Lead with 8 years in B2B SaaS. Data-driven, pipeline-obsessed."
 28  -->
 29  
 30  
 31  TRACK RECORD
 32  <!--
 33  What goes here: 3-5 quantified achievements that demonstrate your level. Team
 34  sizes managed, revenue or scale metrics, products shipped, awards, patents,
 35  certifications. The LLM uses these to calibrate seniority expectations —
 36  vague claims like "led teams" get ignored.
 37  Why it matters: Distinguishes senior from mid-level. Numbers are the signal.
 38  Examples:
 39  - Tech: "Managed team of 8+ engineers. Deployed systems serving 50M+ daily requests. 2 patents, $2M ARR."
 40  - Healthcare: "Led 12-nurse ICU unit across 3 shifts. Reduced patient fall rate by 40%. CCRN certified."
 41  - Marketing: "Managed $2M annual ad budget. Grew qualified pipeline 3x in 6 months. 2 product launches."
 42  -->
 43  
 44  
 45  CORE SKILLS & EXPERTISE (ACTIVE USE, NOT FULL HISTORY)
 46  <!--
 47  What goes here: Skills grouped by category — only what you want to actively use
 48  in your next role. This is NOT your full resume. If you list it here, the LLM
 49  treats it as a core matching criterion.
 50  Why it matters: The LLM matches these against job requirements. Listing everything
 51  dilutes the signal — only include what you want to do daily.
 52  Examples:
 53  - Tech:
 54    - Domain: LLMs, RAG, Information Extraction, Dialogue Systems
 55    - Frameworks: PyTorch, JAX, vLLM, TensorRT-LLM
 56    - Engineering: High-throughput serving, distributed training, Python
 57  - Healthcare:
 58    - Clinical: Critical care, ventilator management, hemodynamic monitoring
 59    - Certifications: CCRN, BLS, ACLS
 60    - Systems: Epic, Cerner, patient safety protocols
 61  - Marketing:
 62    - Channels: Paid search, LinkedIn ads, content marketing, SEO
 63    - Tools: HubSpot, Google Analytics, Looker, dbt
 64    - Methods: A/B testing, attribution modeling, funnel optimization
 65  -->
 66  
 67  
 68  WORKING STYLE
 69  <!--
 70  What goes here: 2-3 bullets describing how you approach problems. This helps the
 71  LLM match culture-fit signals in job descriptions.
 72  Why it matters: Teams that work differently cause friction. This section helps
 73  filter for alignment.
 74  Examples:
 75  - Tech: "Pragmatic & delivery-focused. Prioritizes robust, deployable solutions over academic perfection."
 76  - Healthcare: "Evidence-based practice. Calm under pressure. Advocates for patients with physicians."
 77  - Marketing: "Hypothesis-driven. Runs experiments before committing budget. Prefers small bets over big campaigns."
 78  -->
 79  
 80  
 81  TARGET ROLES & LEVEL (WHAT YOU WANT TO OWN)
 82  <!--
 83  What goes here: What you want to OWN in your next role — not just what you're
 84  qualified for. Seniority level, hands-on vs. management split, and the one
 85  non-negotiable the role must include.
 86  Why it matters: The LLM uses this to distinguish "qualified but unhappy" from
 87  "qualified and motivated." Ownership > capability.
 88  Examples:
 89  - Tech: "Hands-on technical leadership ('playing coach'). Must allow deep involvement in model development and pipeline architecture. Not looking for pure people management."
 90  - Healthcare: "Charge nurse or clinical educator. Must include direct patient care — not looking for desk-only roles."
 91  - Marketing: "Head of Growth or VP Marketing. Must own the full funnel from acquisition to retention — not just brand or just performance."
 92  -->
 93  
 94  
 95  TARGET MARKET (SOFT FILTERS — PENALIZE, DO NOT REJECT)
 96  <!--
 97  What goes here: Industries, company stage/size, geography, and compensation range.
 98  These are soft preferences — the LLM penalizes mismatches but does not auto-reject.
 99  Why it matters: Helps the LLM rank jobs by desirability beyond pure skill match.
100  Examples:
101  - Tech: "Industries: fintech, dev tools, AI-native companies. Stage: Series A-C, 50-300 people. Comp: 90-120K base + equity."
102  - Healthcare: "Settings: academic medical centers, Level I trauma. Size: 200+ beds. Region: Northeast US."
103  - Marketing: "Industries: B2B SaaS, developer tools. Stage: post-PMF, pre-IPO. Comp: 80-100K + variable."
104  -->
105  
106  
107  WORK MODE PREFERENCE
108  <!--
109  What goes here: Remote, hybrid, or onsite preference. The LLM uses this to
110  flag mismatches when a job posting specifies a work mode.
111  Why it matters: A great role with the wrong work mode is still a bad fit.
112  Examples:
113  - "Full remote preferred. Open to hybrid if within commuting distance."
114  - "Hybrid preferred (2-3 days onsite)."
115  - "Remote only — will not consider onsite roles."
116  -->
117  
118  
119  NON-CORE AREAS (DOWNGRADE SCORE, DO NOT AUTO-REJECT)
120  <!--
121  What goes here: Work that drains you but you'd tolerate in an otherwise strong
122  role. The LLM penalizes jobs heavy in these areas but never auto-rejects.
123  Why it matters: Distinguishes "bad fit" from "acceptable tradeoff." Prevents
124  filtering out roles that are 80% great and 20% meh.
125  Examples:
126  - Tech: "Pure web/SaaS without a heavy algorithmic component."
127  - Healthcare: "Administrative duties, committee work, EMR documentation."
128  - Marketing: "Event planning, PR/comms, brand guidelines enforcement."
129  -->
130  
131  
132  RED FLAGS (REJECT ONLY IF CENTRAL FOCUS)
133  <!--
134  What goes here: Dealbreakers if they dominate the role (>80% of daily work).
135  Acceptable as a peripheral part (<20%). The LLM rejects only when these are
136  the core of the job.
137  Why it matters: Prevents false rejections — many jobs mention red flag tasks
138  as minor duties. The centrality threshold matters.
139  Examples:
140  - Tech: "Pure cloud/MLOps (K8s, AWS setup). Low-value API wrapping. 'Black box' roles where you can't touch model internals."
141  - Healthcare: "Pure utilization review. Insurance pre-authorization desk work. Phone triage without patient contact."
142  - Marketing: "Pure social media management. Graphic design. Content writing without strategy ownership."
143  -->
144  
145  
146  DIFFERENTIATORS (WHAT SETS YOU APART)
147  <!--
148  What goes here: What makes you memorable compared to other candidates at your
149  level. Unique skill combinations, unusual career moves, niche expertise.
150  Why it matters: The LLM uses this to identify PERFECT_MATCH opportunities where
151  your specific background is a competitive advantage — not just a fit.
152  Examples:
153  - Tech: "Double background: applied research (patents, publications) + production engineering (50M+ API calls/day). Co-founder experience."
154  - Healthcare: "Bilingual (English/Spanish) ICU nurse with quality improvement certification. Published in peer-reviewed journals."
155  - Marketing: "Former engineer turned marketer — can build dashboards, write SQL, and run campaigns. Rare in marketing leadership."
156  -->