AI Review Authenticity Verification Developer

Intevo AI

Remote · $100k – $130k / year · Full-time · Posted 4 months ago

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

About the role

Fake reviews undermine marketplace trust. A seller with 2,000 reviews, 40% of which are manufactured by a review farm, has an unfair advantage over sellers with genuine feedback — and the buyer who purchases based on those reviews is deceived. FTC enforcement has increased; marketplace liability for hosting fake reviews has grown; and more importantly, buyers who encounter fake reviews lose trust in the entire platform. Our marketplace platform hosts 2.8 million listings and 45 million reviews. Review authenticity is a persistent challenge: review farm operators adapt their generation techniques faster than static rule systems can update. The ML approach needs to detect authentic-seeming manufactured reviews, not just obvious spam. You'll build the review authenticity detection system: training NLP models on stylistic and semantic signals of manufactured content, implementing reviewer behavior analysis that detects coordinated review patterns across accounts, and building the confidence scoring that routes borderline reviews for human moderation rather than automating all decisions. The evaluation framework is critical — precision and recall trade-offs differ between removing a real seller's genuine reviews and letting manufactured reviews stay.

Fully remote.

Responsibilities

Train review authenticity models on stylistic, semantic, and behavioral signals of manufactured content
Implement reviewer behavior analysis detecting coordination patterns across account review histories
Build confidence scoring routing borderline authenticity cases for human moderation review
Develop adversarial robustness as review farm techniques evolve toward more authentic-seeming content
Create precision-recall trade-off analysis supporting policy decisions on moderation thresholds
Monitor false positive rates ensuring authentic reviews aren't removed at unacceptable rates

Requirements

3+ years in NLP engineering with content moderation, fraud detection, or trust and safety experience
Strong Python and PyTorch skills with text classification model development
Understanding of adversarial content generation and review manipulation techniques
Experience building human-in-the-loop moderation workflows for consequential decisions
Ability to navigate precision-recall trade-offs in policy-sensitive detection contexts