Conversational AI Designer, full-time, fully remote, with compensation reaching $100,000 a year.
The basics
This is a full-time position with no office tied to it, open to candidates in any location. Pay tops out at $100,000 annually, with the exact amount depending on experience and interview outcomes. There's no relocation requirement built into this role, and the working hours flex around a small daily overlap window for reviews and cross-team check-ins.
What you'd actually design
The work centers on building the conversations that sit inside chatbots and voice assistants, from the first line of a script through the edge cases that trip up most flows. Consider a customer typing that their order arrived damaged. A poorly designed flow might respond with the same upbeat, scripted cheerfulness it uses for a simple shipping question, which reads as tone-deaf the moment someone's actually frustrated. Getting that tone shift right, consistently, across dozens of similar edge cases, is a real part of the job, and it's rarely something a single pass at scripting catches on its own. It usually takes a second or third round of testing, with a fresh set of eyes, before those mismatches surface clearly.
- Script and structure dialogue flows for chatbots and voice assistants
- Test conversations for clarity and tone across a range of user inputs, not just the happy path
- Partner with engineers to refine natural language understanding and the resulting user experience
A good flow anticipates the messy, half-finished way people actually type or speak, not the clean version a script outline assumes. Someone canceling a subscription rarely phrases it the same way twice, and a flow built around only the most obvious phrasing breaks the moment a real user goes off-script.
Skills this role leans on
Conversation design experience is the core requirement, paired with a working understanding of natural language understanding, since a designer here needs to know roughly what a model can and can't reliably parse before writing around it. UX writing matters just as much as the flow logic itself; a technically correct response that reads awkwardly still fails the user. Beyond that, comfort with chatbot platforms, prototyping tools, structured user testing, and dialogue flow mapping all come up regularly. Prototyping tools, in particular, are used earlier in the process than people expect, often before a single line of production script is written, just to see whether a proposed flow even makes sense out loud.
None of these need to be mastered equally. Someone strong on the writing side with lighter platform experience is a normal starting point, and the reverse is just as workable, as long as the fundamentals of how a conversation actually breaks down are solid. A designer who can spot why a flow feels stilted, even without being able to name the exact platform limitation causing it, usually picks up the technical side quickly once they're working inside a real system.
Background that fits
A bachelor's degree is the baseline expectation for this role, most often earned in a language-, design-, or communication-focused program rather than a purely technical one, since the job leans heavily on how people actually talk. Naukri Mitra has seen strong candidates arrive from less obvious academic backgrounds too, provided they can point to real chatbot or voice assistant design work. A year and a half of hands-on experience shaping these kinds of interactions is the minimum bar. Familiarity with at least one conversational AI platform is expected going in, rather than something picked up entirely on the job.
What separates a strong application from a mediocre one is usually evidence of actual shipped conversations, not just coursework or a portfolio built entirely around hypothetical scripts. A single well-documented flow that went through real user testing and got reworked as a result says more than a stack of polished but untested examples.
What comes with the role
Full-time status brings a standard package here, built around a few things specific to remote, design-heavy work. None of these are treated as optional extras layered on top of a bare-bones offer; they're built into the role's structure from the start.
- Health coverage
- Paid time off
- Genuine remote-work flexibility, without a fixed core-hours requirement beyond occasional overlap for reviews
- A professional development budget that can go toward UX training or AI-focused courses and conferences, whichever fits a designer's own gaps better
How the team works
Design and engineering sit close together here, which means a dialogue flow rarely gets finalized in isolation before an engineer weighs in on what's actually feasible to build. That back-and-forth can slow a script down in the short term, but it usually catches problems long before a flow reaches real users. Expect regular testing sessions where a designer walks through a flow out loud, catching the moments where phrasing sounds fine on paper but stumbles the instant it's spoken or typed by someone unfamiliar with the product. These sessions tend to surface more real issues than any amount of desk-based solo review.
Voice interfaces in particular expose problems that text interfaces hide. A phrase that reads perfectly clear on a screen can sound clumsy or ambiguous once it's spoken aloud, so a fair amount of the testing work happens with actual audio, not just a written script. Something as small as where a sentence pauses can change how a user interprets an instruction, and that kind of detail almost never shows up until someone actually hears it read back.
Applying
Send a resume along with examples of dialogue flows or scripts you've built, ideally including a case where a flow needed real revision after testing revealed a problem. Applications are reviewed on a rolling basis, and there's no fixed deadline for this posting. Candidates who include a specific example of a conversation they redesigned after it failed in testing tend to stand out more than those who only describe their process in general terms. A short writing sample from a real project, even an unpolished one, is usually more useful here than a long list of tools listed without context.
