About The Project
This project delivers a
Thai Voice AI PoC that automates high-volume outbound telemarketing calls using real-time voice interaction and LLM-based intent handling. The solution improves call consistency and conversion quality, benchmarks multiple ASR/TTS providers in Thailand, and enables safe human handoff for complex cases. The PoC provides ILA with clear performance metrics and a scalable foundation for expanding voice automation without linear cost growth.
Duration: 3-5 weeks
KPI
KR1: 95% test coverage for top telemarketing scenarios and top 20 intents
KR2: 0 P0 / P1 defects open at PoC demo and readout
KR3: Human handoff triggers correctly in 99% of low-confidence or edge cases
KR4: Consent / opt-out / fallback flows pass 100% of compliance test cases
KR5: Escalation and hang-up logic validated across all supported call paths
Areas of Responsibility
- Define and maintain QA strategy for Voice AI PoC MVP
- Design test scenarios for telemarketing flows, intents, and edge cases
- Validate ASR/TTS quality (WER/CER, accent robustness, noise conditions)
- Test LLM intent accuracy, fallback behavior, and confidence thresholds
- Verify human handoff, consent, DNC, and compliance-related logic
- Analyze QA metrics and provide go/no-go recommendations
- Ensure demo and readout readiness (no surprises on demo day)
Skills
- QA for voice, conversational AI, or IVR systems (Thai language)
- Scenario-based and exploratory testing
- Experience with API, backend, and integration testing
- Log analysis, trace inspection, and defect root-cause analysis
- Ability to translate business flows into testable acceptance criteria
Knowledge
- Voice AI fundamentals: ASR, TTS, WER, CER, latency metrics
- LLM-based intent detection and confidence thresholds
- Call flow logic: barge-in, interruptions, silence handling, hang-ups
- Telemarketing and contact-center workflows (nice to have)
- QA metrics and release readiness criteria
- Basic understanding of compliance concepts (consent, opt-out, escalation)
