Remote AI Benchmark Engineer & Multilingual NLP Specialist

LILT; United States; Remote; USD 83,000 - 165,000; Part time

Benefits offered by this job

Flexible schedule

Competitive freelance rates

Global project exposure

Autonomy in project choices

Diverse language-focused work

Remote collaboration

Job summary

LILT seeks experienced native-speaking software engineers to design, build, and validate multilingual benchmarks. You will create high-signal tasks and assets that stay in the target language to test multilingual handling.

This is a remote, freelance opportunity. You will perform task engineering, asset creation, prompting, verification, and calibration across model tiers, with a 4-layer human and automated QA process.

Qualifications

• 5+ years of industry software engineering experience.
• Proven track record at leading technology companies.
• Native or near-native fluency in the target language with high English proficiency.

Responsibilities

• Task Engineering: Evaluate Coding Agents.
• Asset Creation: Build datasets and files in your native language, kept in the target language.
• Prompting & Translation: identify failure points in your native language.
• Implementation & Verification: develop reference implementations and deterministic verifier scripts.
• Calibration & Execution: analyze logs and calibrate task difficulty across model tiers.
• Quality Assurance: participate in a four-layer quality control process with automated checks.

Skills

Python

Shell scripting

Data processing

Terminal/CLI workflows

Multilingual text processing

English proficiency

Education

Bachelor's degree in computer science or related field

Job description

LILT seeks experienced native-speaking software engineers to design, build, and validate multilingual benchmarks. You will create high-signal tasks and assets that stay in the target language to test multilingual handling.

This is a remote, freelance opportunity. You will perform task engineering, asset creation, prompting, verification, and calibration across model tiers, with a 4-layer human and automated QA process.