Anyone Have Experience Working with an NLP Development Company Recently

[RemoteState]

Throwing this out there because we are about to make a decision and could really use some outside perspective.

We are trying to hire nlp developers for a project involving customer feedback analysis across multiple languages, not just English. Sounds straightforward until you realize most nlp development services providers we have talked to have only ever worked with clean English language datasets. The moment we mentioned multilingual support and regional dialects half the conversations got noticeably vaguer.

We shortlisted two companies so far. Both look solid on paper. Both have decent case studies. But paper only tells you so much and I would rather hear from someone who has actually gone through this before we commit budget.

What We Are Trying to Verify Before Moving Forward​

Put together a short list of things to dig into with both companies based on advice from a few people who have hired NLP teams before. Sharing here in case it helps someone else in the same spot:

• Asked each nlp development company to walk us through a past project involving messy or non-English data because that reveals more than any polished case study ever could
• When you hire nlp developers, confirm whether they are strong in classical NLP techniques or leaning entirely on large language models because the right approach really depends on your specific use case and budget
• Checked if the nlp development services included proper evaluation benchmarks tied to business outcomes not just accuracy scores that sound impressive but mean nothing to our team
• Asked how they handle domain specific vocabulary since generic models often completely misread industry jargon or slang common in customer feedback
• Wanted to know if the same developers who scoped the project would actually hire nlp developers internally to build it or if it gets handed off to a completely different team
• Confirmed ongoing model maintenance was included since language and slang shift constantly and a static model degrades faster than most clients expect

Still Deciding​

Genuinely torn between the two options right now. If anyone here has worked with an nlp development company recently, especially for multilingual or industry specific use cases, I would really appreciate hearing how it went. Good or bad experiences both welcome at this point, just trying to make an informed call.