Company: Voxket.ai

Location:
Greater Bengaluru Area, India Employment: Full-time | Hybrid Voxket.ai is building the next generation of AI employees that autonomously execute frontline and back-office operations across voice, WhatsApp, email, and web channels. Our AI systems combine real-time conversational intelligence, speech processing, document understanding, and workflow automation to deliver reliable, production-grade AI solutions. We're building our own AI capabilities, optimizing models for real-time performance, and developing scalable inference infrastructure to power enterprise AI deployments. Role: We're looking for passionate AI/ML Engineers with solid experience in deep learning, speech and audio AI, model training, and high-performance inference. You will work on developing, fine-tuning, evaluating, and deploying AI models, with a particular focus on Automatic Speech Recognition (ASR), Text-to-Speech (TTS), multilingual conversational AI, and real-time audio processing. The role involves working across the ML lifecycle, from dataset preparation and model experimentation to GPU optimization, distributed training, and scalable production inference. You will collaborate closely with our engineering and research teams to build AI systems capable of delivering low-latency, high-quality interactions at enterprise scale. Key responsibility
• Design, train, fine-tune, and evaluate deep learning models for speech recognition, speech synthesis, and conversational AI.
• Develop and optimize multilingual ASR/STT and TTS models, including Indian languages and code-switching scenarios.




• Build scalable, low-latency inference pipelines for real-time audio streaming and conversational applications.
• Optimize model performance through quantization, distillation, batching, and GPU acceleration.
• Work with distributed training frameworks and GPU infrastructure to train and serve large-scale models.
• Develop robust audio data pipelines, including preprocessing, augmentation, segmentation, and dataset quality validation.
• Benchmark and improve model accuracy, inference latency, throughput, and GPU utilization.
• Deploy and monitor models in production environments, ensuring reliability, scalability, and cost efficiency.
• Collaborate on model architecture, experimentation, and research-driven improvements. Required Qualifications
• Strong understanding of machine learning, deep learning, neural networks, and Transformer architectures.
• Proficiency in Python and PyTorch, with hands-on experience training and fine-tuning deep learning models.
• Experience with speech and audio processing, including ASR, TTS, speech enhancement, or related audio ML applications.
• Familiarity with speech architectures such as Conformer, CTC, RNN-T,



encoder-decoder Transformers, or neural vocoders.
• Experience with GPU-based inference optimization, Good to Have
• Experience with models such as Whisper, wav2vec 2.0, HuBERT, VITS, FastSpeech, XTTS, or similar architectures.
• Familiarity with NVIDIA Triton Inference Server, TensorRT, ONNX Runtime, or TorchServe.
• Experience with CUDA, GPU profiling, kernel optimization, or memory-efficient inference.
• Knowledge of DDP, FSDP, DeepSpeed, or other distributed training frameworks.
• Experience with streaming ASR, low-latency TTS, voice activity detection (VAD), and speaker diarization.
• Familiarity with Docker, Kubernetes, autoscaling, and cloud GPU infrastructure.
• Experience with multilingual speech datasets, Indian language models, and code-switching.
• Understanding of real-time factor (RTF), word error rate (WER), character error rate (CER), and speech quality evaluation. Why Join Voxket.ai? You'll have the opportunity to work on challenging, real-world AI problems, build and optimize speech models, and develop inference systems that serve enterprise customers. We're looking for engineers who enjoy going beyond API integrations — people who love understanding how models work, improving their performance, and taking them from experiments to reliable production systems. If you're passionate about speech AI, deep learning, and high-performance inference, we'd love to hear from you. Apply or reach out to us directly to explore the opportunity.

📌 Artificial Intelligence And Machine Learning Engineer (Bengaluru)
🏢 VoxKet.AI
📍 Bengaluru