Actively recruiting / 190 applicants
We’re here to help you
Cynthia is in direct contact with the company and can answer any questions you may have.
Cynthia, RecruiterFull-Stack Voice AI Engineer
Why This Opportunity
Help an early-stage voice AI product work reliably in real conversations. This part-time contract focuses on the engineering foundations that become important when a phone-based AI system reaches real users.
About the Company
A company is developing an early-stage phone-based voice AI product. An existing MVP is in place, and the team wants practical engineering input to improve its reliability and scalability.
Why This Role Matters
A working demo is only the beginning. Longer conversations, response delays and unexpected model behavior create challenges that need thoughtful architecture and testing. Your work will help strengthen the product's foundations before further development.
What You'll Build
- Improve prompt architecture so the agent follows instructions consistently throughout a conversation.
- Define how deterministic guards, tools and the LLM work together to control behavior.
- Improve how call state is tracked and used across the voice pipeline.
- Identify opportunities to reduce latency and improve conversational turn handling.
- Design a testing approach that complements the existing unit tests with backtesting against real or synthesized call data.
- Evaluate both LLM logic and audio behavior, including voice activity detection (VAD) and turn detection.
- Contribute to the relevant Python and full-stack product components needed to support these improvements.
What Success Looks Like
The initial work should establish stronger foundations for prompt design, deterministic controls, call-state management and repeatable testing. The emphasis is on practical improvements informed by how voice systems behave with real users.
Who We're Looking For
Must Have
- Hands-on Python development experience, with contributions across relevant backend, integration or full-stack product components.
- Experience developing or improving voice AI systems, including the interaction between speech services, an LLM and application logic.
- Practical knowledge of prompt engineering, tool interactions and state management for conversational applications.
- The ability to investigate reliability or latency problems and help design tests for LLM-driven or voice applications.
- Based in the United States or Europe, excluding Turkey, subject to applicable engagement restrictions.
Nice to Have
- Experience operating voice AI with real users beyond a demo or an MVP handoff. This is a particularly valuable differentiator for the role.
- Familiarity with Twilio, Speechmatics or ElevenLabs. Experience with every vendor is not required.
- Experience with deterministic guardrails, call-state tracking, or latency improvements in production voice systems.
- Experience backtesting changes against recorded or synthesized calls, including VAD or turn-detection evaluation.
- Familiarity with early-stage product development and prioritizing improvements within a limited weekly time budget.
Tools & Technologies
Application Development
- Python, with relevant full-stack and integration components.
Voice AI Pipeline
- Twilio, Speechmatics, gpt-5.4-mini and ElevenLabs form the current cascade stack.
Quality & Evaluation
- Existing unit tests, with a focus on stronger backtesting of LLM logic and audio behavior.
Compensation & Benefits
- Approximately 10 hours per week for an estimated 8 weeks initially.
- Potential for follow-on work if the initial engagement goes well; no extension is guaranteed.
Ready to Apply?
If this work matches your experience, we'd like to hear about a voice AI system you've helped build or improve and the problems you personally solved. You do not need every preferred qualification to apply.