RocketHuntSmart job search
6 months ago
3,500 – 5,000 USD
per month
Reach out directly about this role
Grade
Middle
Experience
from 2 years
Work Format
Hybrid, Remote
RocketHunt estimate from real salaries
AI writes it from your resume. You only have to send it.
Company: International VoIP / Cloud Telephony product Format: hybrid (Tashkent 2–3 months → then remote) Salary: $3500–5000 (negotiable)
They are building a high-load cloud telephony platform for the USA/Canada, where ML is a core part of the system. They are looking for an ML engineer who knows how to bring speech/audio models to production and be responsible for their real-time operation.
Responsibilities: • Development of AMD (Answering Machine Detection) in real-time calls • End-to-end ML: from audio labeling to deployment and calibration in production • Integration of models into the backend (C# / SIP / RTP) via ONNX Runtime • Latency optimization (audio streaming, working with milliseconds) • Analysis of edge cases and errors in real calls R&D: VAD, noise suppression, speech models
Tech Stack: Python, C#, wav2vec 2.0, Whisper, HuggingFace, MFCC, spectrograms, ONNX Runtime, quantization, SIP/RTP
Requirements: • 2+ years of ML in production • Experience with Speech / Audio (ASR, VAD, audio classification) • Ability to work end-to-end (data → inference) • Understanding of metrics and quality (Precision/Recall, ROC-AUC, calibration) • Engineering approach and production work, not just experiments
Conditions: • Offline onboarding in Tashkent (2–3 months), then remote • High-load production ML • Strong engineering team, minimal bureaucracy
📩 CV to Telegram: recruiter contact
