ML Engineer – ASR

This is a fully remote position, open to applicants in Greece.

📋 Description

• Enhance automatic speech recognition (ASR) for Sophea Meet.

• Manage the improvement cycle from data and model experiments to evaluation and production performance.

• Train, fine-tune, adapt, and assess speech recognition models according to identified product requirements.

• Create representative benchmarks and regression tests for actual meeting audio.

• Prepare speech datasets and analyze challenging scenarios, such as noise, overlapping speakers, extended sessions, and specialized terminology.

• Conduct error analysis to identify which changes in models, data, or pipelines enhance transcription quality.

• Boost inference accuracy, latency, throughput, reliability, and cost-effectiveness.

• Deploy and oversee model versions in collaboration with Engineering.

• Collaborate with Product to establish quality objectives and prioritize enhancements.

• Prioritize additional language support, starting with Greek, based on usage and product demands.

⛳️ Requirements

• Over 3 years of relevant experience in ML or research engineering, demonstrating clear technical leadership and measurable outcomes.

• Strong hands-on experience with ASR or speech ML, including a speech recognition system that you have launched or significantly improved.

• Proficient in training, fine-tuning, or adapting deep learning models utilizing real-world audio data.

• Excellent evaluation skills, encompassing WER/CER, benchmark design, error analysis, and regression testing.

• Proficient in Python programming.

• Practical experience with PyTorch, JAX, TensorFlow, or similar frameworks.

• Solid software engineering background.

• Experience in deploying, monitoring, and maintaining production ML systems.

• Experience in multilingual ASR, accent robustness, domain adaptation, or diarization is a plus.

• Familiarity with streaming speech recognition and low-latency inference is a plus.

• Experience with MLOps, evaluation, or data pipelines for speech systems is a plus.

• Background in meeting transcription, voice agents, or enterprise speech products is a plus.

🏝️ Benefits

• Competitive compensation package aligned with industry benchmarks.

• Hands-on opportunities to shape AI capabilities for Kiefer Tech’s products and real-world applications.

• AI-native environment with challenges in model development, evaluation, and production AI systems.

• Engineering-first culture.

• High degree of autonomy.

• Minimal bureaucracy.

• Opportunity to contribute to meaningful AI product development.