Post a Job
All remote jobs
Welo Global

Project Lion – Senior Prompt Engineer – Japan (Remote, Part-Time)

AI Summary
This part-time freelance role leads the technical migration of evaluation templates to LLM-based autoraters for a Japan-based client. The Senior Prompt Engineer will generate, optimize, test, and manually refine prompts for complex structured classification workflows. Key work includes monitoring human-versus-model disagreements and evaluating quality through F1, precision, recall, and gold-data testing. The position requires native Japanese, fluent English, substantial prompt-engineering experience, and strong analytical independence. It also involves preparing technical launch-readiness documentation for final deployments.
Apply for this job View company
Remote from
Japan
Salary
Undisclosed
Department
Data Science & Analytics
Employment
Contract
Experience
Senior
Published
Apply before
8 Dec 2026
Listing views
28
Application actions
1
Application toolkit

Make your next move.

Prepare your resume, explore your fit, and draft a cover letter for this opportunity.

Opportunity details

About this role.

 We are seeking a Prompt Engineer to be responsible for the end-to-end technical migration workflow for transitioning templates to LLM autoraters. The role is required to use client’s internal tools to leverage prompt engineering techniques to maximize model performance.

Responsibilities:

  • Utilize Automatic Prompt Generation (APG) tools to create baseline prompts for complex parent-child template clusters.

  • Run and supervise Automated Prompt Optimization (APO) tool, review the outputs, and flag when the APO reaches deadlocks or plateaus.  

  • Manually draft, test, and refine prompts to navigate complex template architectures, overcome anti-patterns, and handle edge cases where tooling is lacking or broken. Solve edge-case scenarios by designing and refining manual prompts.

  • Monitor shadowbot runs to ensure sufficient disagreements (between human and LLM ratings) are registered, generated, and tracked.  

  • Run prompt versions against established gold data to continuously measure autorater quality against the human crowd baseline, calculating accuracy metrics such as F1 scores, precision, and recall.  

  • Draft technical launch readiness justifications (Launch Certification Documentation) for final.

Requirement:

  • Language Skills: Native fluency in Japanese and fluent in English.

  • Location: Must be based in Japan.

  • Education: Bachelor’s, Master’s, or Doctorate degree in Computer Science, Data Science, Computational Linguistics, Human-Computer Interaction (HCI), Cognitive Science, or a related analytical field. 

  • Prompt Engineering & AI Expertise: At least 4 years’ experience as Prompt Engineer. Proven experience tuning Large Language Models (LLMs) for strict, structured outputs, complex classification tasks, and familiarity with chain-of-thought and few-shot learning. 

  • Data Analysis: Strong proficiency in identifying error patterns, analyzing model performance, and using SQL or other data analytics tools. 

  • Technical Agility: Ability to quickly learn and master proprietary tools with minimal supervision. 

  • Communication: Excellent verbal and written communication skills. 

Optional / Preferred Skills: 

  • Familiarity with enterprise-grade LLM interfaces like the Goose API. 

  • Experience in AI model evaluation, data science, computational linguistics, or software engineering. 

  • Hands-on experience with Automated Prompt Optimization (APO) systems or tuning workflows. 

  • Linguistic expertise, including an understanding of semantics and logic. 

Compensation

Additional Information

Apply now >

This job listing has been manually reviewed by the Jobicy Trust & Safety Team for compliance with our posting guidelines, including verification of the company's legitimacy, accuracy of job details, clarity of remote work policy, and absence of misleading or fraudulent content.

Role DNA

A quick view of the complexity, pace, ownership and collaboration implied by the job description.

Job Complexity

5/5
EasyHard

Pace & Pressure

4/5
RelaxedFast-paced

Autonomy Level

5/5
GuidedFull ownership

Communication Load

4/5
IndependentCollaborative
AI insightThe role combines advanced LLM prompt optimization, model-evaluation methodology, Japanese-language expertise, and troubleshooting of complex edge cases. Success depends on independently diagnosing quality failures and producing defensible launch documentation.

Salary analysis

Estimated compensation compared with the broader US market for similar roles.

Estimated job median (per hour)Market rate
$55
US market range (per hour)$40–$75
AI insightThe hourly pay period is inferred from the job context; it was not explicitly stated by the employer. No employer compensation is disclosed. The hourly cadence is inferred from the explicitly part-time freelance engagement; estimated USD hourly market range for a senior Japan-based prompt-engineering and LLM-evaluation contractor is $40–$75, with an estimated midpoint of $55.

Core skills

Skills and capabilities most closely associated with this opportunity.

Prompt engineeringLarge language modelsLLM evaluationAutomated prompt optimizationAutomatic prompt generationPrompt testingModel performance analysisF1 score analysisPrecision and recallSQLJapanese language proficiencyEnglish communicationComputational linguisticsTechnical documentationError pattern analysis
Sample interview questions
How would you evaluate whether an optimized prompt is ready for production use?

I would test it against representative gold data, compare results with the human baseline, and measure precision, recall, F1, and error distribution. I would also review failure modes qualitatively, particularly for edge cases and strict output-format compliance, before documenting the launch recommendation.

What would you do if an Automated Prompt Optimization workflow reaches a plateau?

I would inspect the prompt variants, evaluation set composition, and metric-level errors to determine whether the limitation comes from the prompt, data, or task definition. I would then form targeted manual hypotheses, such as clarifying label boundaries, adding carefully selected few-shot examples, or decomposing ambiguous instructions.

How do you design prompts for strict structured outputs?

I define an explicit schema, unambiguous label definitions, formatting constraints, and decision rules for uncertain cases. I validate the output with adversarial and edge-case examples, then iterate using observed parsing failures and classification errors.

How would you investigate persistent disagreement between human raters and an LLM autorater?

I would segment disagreements by label, template, linguistic feature, and confidence pattern, then review examples to identify unclear instructions, inconsistent human judgments, or systematic model bias. Based on the findings, I would refine the rubric or prompt and verify the improvement on held-out gold data.

How would your Japanese-language expertise improve an LLM evaluation workflow?

I would assess whether prompt instructions and examples correctly capture Japanese semantics, ambiguity, politeness levels, context dependence, and domain-specific phrasing. I would use this analysis to create more representative test cases and reduce errors caused by literal or culturally incomplete interpretations.

This analysis is generated from the job description. Salary estimates, role characteristics and sample answers are guidance, not employer-provided facts.
Next step

Apply now.

Follow the employer’s application method and review Jobicy’s safety guidance before sharing personal information.

Did you apply?Let us know, and we’ll help you track your application.

Continue on the employer website

Protect your personal information and never pay to secure an interview or job offer. .

Log in to save
One quick step before you apply

Sign in to continue.

Sign in or create a free account to continue to the employer's application.
Add alert
Job Search Safety Tips
Here are some tips to help you search and apply for jobs safely:
Watch out for suspicious jobs Don't apply for jobs that offer high pay for little work or offer to hire you without an interview. Read more ›
Check the employer's profile Make sure you're applying for a trustworthy job by visiting the employer's profile and learning more about them. Read more ›
Protect your information Don't share personal details like your bank account or government-issued ID on suspicious websites or messengers. Read more ›
Report jobs that feel unsafe If you see a job that seems misleading, inappropriate or discriminatory, report it for going against our policies and we'll review it.
Share this job
Jobs Talent AI Tools Salaries
Menu