Senior Ai Engineer (STT, Ai/ML exp) with Japan based IT Product based startup Co - Bengalore

Position- Sr Ai Engineer

Experience- 4+years

Work mode:-5 days Work from office

Location-Bengaluru

Key Responsibilities

• Design, build, and optimize speech-to-text (STT) pipelines for physician-facing product features.
• Fine-tune, evaluate, and deploy open-source LLMs and other ML models for production use.
• Design and build recommendation engines to improve physician engagement and content relevance.
• Work across the full ML lifecycle - data collection/curation, training, evaluation, deployment, and monitoring.
• Collaborate closely with backend engineers to integrate models into production APIs and services.
• Continuously evaluate new open-source models, techniques, and tools relevant to STT, recommendations, and LLM fine-tuning.
• Optimize models and pipelines for latency, cost, and accuracy trade-offs appropriate for a production medical platform.
• Build evaluation frameworks and benchmarks to measure model quality and catch regressions before they reach production.
• Work with vector databases, retrieval pipelines, and search infrastructure where relevant to recommendation and search features.
• Investigate and resolve production issues related to model performance, latency, or output quality.
• Document experiments, model decisions, and architecture choices clearly for the team.

Skills needed

• 3+ years of hands-on experience building and deploying ML/AI systems in production.
• Experience with STT (speech-to-text) systems - building, fine-tuning, or integrating them.
• Experience training or fine-tuning open-source models (LLMs or other architectures).
• Experience building recommendation systems or ranking/personalization models.
• Solid understanding of the ML lifecycle: data pipelines, training, evaluation, deployment, monitoring.
• Comfortable working with Python and standard ML/DL frameworks (PyTorch, Hugging Face, etc.).
• Ability to independently investigate unfamiliar technical problems and identify practical solutions.
• Strong sense of ownership - comfortable taking a model from idea to production, not just notebook experiments.