I’m building a real-time, two-way translator dedicated to business calls and need an expert who can take it from concept to a working product. The first release must handle English ⇄ Spanish flawlessly, converting speech to text, translating it, then rendering clear synthesized speech back to both parties with minimal latency. Compatibility is non-negotiable: the same core engine has to run inside a web browser, a mobile app (iOS and Android), and a lightweight desktop client. I value reusable backend services—WebRTC for voice transport, a robust ASR + NMT pipeline (DeepSpeech, Whisper, or similar paired with a proven translation model), and near real-time TTS. Security, call recording toggles, and an admin dashboard for basic analytics should round out the feature set. If you’ve previously shipped AI voice solutions or low-latency streaming apps and can demonstrate sub-800 ms round-trip translation, I’d like to see your approach: preferred stack, model choices, and any optimisation strategies for scaling concurrent calls. Future phases may expand to French or Chinese, so designing with multilingual extensibility in mind will be appreciated. Please outline the milestones you foresee—from prototype to production deployment—and link to any live demos or repos that showcase similar work.