About the role
You build the voice AI agents we ship to clients, and you own them from the inside — the conversation flow, the prompts, the integrations, the behavior. When an agent goes live, you are accountable for how well it holds a conversation, completes the task, and stays inside scope. You turn an approved scope of work into a working agent that a real customer talks to and trusts.
What you'll own
Agent build & configuration
- Build and configure voice agents on VersaTalk and the underlying platforms we deploy on (Retell conversation flows, and contact-center platforms such as NICE CXone where applicable).
- Translate client requirements and approved scope into working conversation flows — nodes, branches, global nodes, edge conditions, and handoffs.
- Set up the integrations an agent needs to do its job: APIs, webhooks, function calls, and knowledge bases.
- Coordinate telephony and routing configuration with the platform/infrastructure team (SBC, SIP) at the configuration level.
Prompt engineering & dialogue
- Write, structure, and version system prompts, node prompts, and dialogue copy.
- Tune prompts for accuracy, natural tone, brand voice (American English for public-facing copy), latency, and cost — including good prompt-caching hygiene to keep per-minute cost where it should be.
- Design fallbacks, escalation paths, and guardrails so the agent fails safely.
Testing & iteration
- Run test calls, debug import errors, fix loops and dead-ends, and read call logs and transcripts to find root causes.
- Iterate fast on the back of QA findings and evaluation results from the Data & Analytics analyst.
Performance ownership
- Own the build-side performance of every agent you ship: containment, task completion, accuracy, latency, and cost-per-minute.
- Keep each agent's configuration and context documentation current as scope changes.
What a strong first 90 days looks like
- You can independently take a scoped use case and stand up a working agent in a test environment.
- You can diagnose a failing call from the logs and ship a fix without escalating.
- You've taken ownership of at least one live agent's configuration end to end.
Required
- Hands-on experience building conversational or voice agents (Retell, Vapi, Bland, Voiceflow, or similar) — or strong prompt-engineering ability plus the appetite to learn a new platform quickly.
- Excellent written English and a genuine ear for natural, on-brand dialogue.
- Comfortable reading logs, debugging, and working with basic API / webhook concepts.
- Methodical, detail-oriented, and organized — small configuration mistakes have real consequences in production.
Nice to have
- Telephony or SIP familiarity.
- Experience optimizing LLM cost and latency.
- Familiarity with contact-center platforms (NICE CXone, Genesys, etc.).
- Working knowledge of Python or JavaScript.
Tools you'll work with
VersaTalk · Retell · NICE CXone / DEVone · LLM APIs (Anthropic, OpenAI) · speech-to-text (Deepgram) · Git.
Reporting line
Reports to the Senior Architect, working closely with the delivery lead.
Fill below form to apply:(Only forms filled applications will be considered)
https://docs.google.com/forms/d/e/1FAIpQLSfOpFj_bIR4lGpNNVlXKOZSNRoAYSEcHSsrcUo4_RDPreAdJQ/viewform?usp=dialog
Pay: ₹400,000.00 - ₹500,000.00 per year
Work Location: Remote
