Sr. Machine Learning Engineer (Speech) - AI

📋 Description

• • Own and evolve the speech-to-text backbone that powers Level AI’s flagship contact-center intelligence platform. You will architect, train, and scale automatic speech recognition (ASR) pipelines that process millions of real-time customer calls every week, turning raw audio into high-accuracy transcripts that fuel downstream analytics, sentiment detection, and agent coaching.
• • Lead end-to-end experimentation on cutting-edge ASR challenges—transcription, speaker diarization, voice-activity detection, noise robustness, and multilingual adaptation—translating research breakthroughs into production-grade services with sub-second latency and 99.9 % uptime.
• • Collaborate daily with product managers, full-stack engineers, and data scientists to design modular, cloud-native architectures that can be re-configured per client in minutes. Your code will define the interfaces between audio ingestion, GPU-accelerated inference, and downstream NLP micro-services, ensuring seamless data flow from microphone to insight dashboard.
• • Continuously benchmark against state-of-the-art open-source and proprietary models (Whisper, wav2vec 2.0, Conformer, RNN-T). You will run rigorous A/B tests on model architectures, training curricula, data augmentation, and decoding strategies, publishing internal tech notes that influence Level AI’s quarterly roadmap.
• • Build and maintain automated MLOps workflows for data labeling, model training, canary releases, and rollback. You will integrate monitoring, alerting, and auto-scaling policies so that every new model version improves both accuracy and cost-efficiency without manual intervention.
• • Mentor junior ML engineers and interns through code reviews, pair programming, and weekly knowledge-sharing sessions. Your guidance will raise the bar for scientific rigor, software craftsmanship, and responsible AI practices across the broader AI org.
• • Partner with customer-facing teams to translate real-world audio pain points—regional accents, low-bandwidth calls, overlapping speakers—into prioritized research sprints. You will present findings to Fortune 500 clients, demonstrating how Level AI’s speech engine outperforms legacy vendors by double-digit WER reductions.
• • Stay ahead of the curve by attending top-tier speech conferences (Interspeech, ICASSP), maintaining an active publication record, and filing patents that secure Level AI’s technological edge. Your thought leadership will shape the next generation of AI-native contact-center products.
• • Work in a fast-moving, agile squad that ships to production weekly. You will participate in sprint planning, daily stand-ups, and retros, ensuring that every commit advances both user value and scientific excellence.
• • Thrive in a hybrid work culture from our Noida tech hub, collaborating synchronously with Mountain View counterparts and asynchronously across global time zones. Your contributions will directly influence product strategy and company valuation as we scale toward IPO readiness.