About GreyLabs AI GreyLabs AI is an Agentic AI company transforming how businesses interact with their customers through Voice AI and Conversational AI. We build AI solutions for enterprise customers, particularly in Banking, Financial Services and Insurance (BFSI), where accuracy, reliability, latency and cost are critical. Our ambition is to build AI models that power conversations touching billions of lives every day. We are focused on developing intelligent, reliable and efficient models that deliver exceptional quality while operating at massive scale. The Opportunity We are looking for an exceptional AI researcher and engineer to shape and build GreyLabs’ next generation of speech and language models. This is a senior, deeply technical role for someone with substantial hands-on model-building experience who wants to own a major AI charter at a high-growth startup. We value demonstrated technical depth and shipped model outcomes over tenure alone. You will work across speech recognition, language models, model training and adaptation, and inference optimisation. You will define technical direction, solve our hardest modelling challenges, and translate research into production systems powering real-world Voice AI applications. This is a builder-leader role, not a management-only position. You should be comfortable writing code, designing and running experiments, debugging model failures, fine-tuning models and optimising inference yourself. You will set direction while remaining directly involved in the hardest technical problems, and mentor researchers and engineers as the team grows. You will own the end-to-end speech and language model charter—from research strategy and model development to production deployment and continuous improvement. You will make build-versus-adapt decisions and be accountable for model quality, inference economics and production performance. The mission is to build models that deliver the right balance of quality, scale, latency and cost, creating differentiated AI capabilities for Indian languages and enterprise conversations. What You Will Do 1. Build and Advance Speech Models
• Develop, train, fine-tune and improve Automatic Speech Recognition (ASR) models for Indian English, Hindi, code-mixed speech, regional languages, diverse accents and BFSI terminology.
• Solve real-world challenges including noisy telephony audio, Indian names, numbers, dates, financial terminology and other business-critical entities.
• Drive model architecture, training, domain adaptation, data strategy and evaluation to improve performance on real-world conversations.
• Explore emerging speech architectures, training techniques and modelling approaches to push the boundaries of recognition quality and efficiency. 2. Develop Language Models for Conversational AI
• Research, adapt and optimise open-weight LLMs for real-time voice conversations and enterprise use cases.
• Improve instruction following, tool calling, conversational consistency, task completion and adherence to business rules.




• Evaluate emerging models and training techniques, making informed decisions about fine-tuning, post-training, distillation and training from scratch where justified.
• Develop systematic approaches to identify and address hallucinations, reasoning failures and inconsistent model behaviour. 3. Build for Scale, Quality and Cost
• Design and optimise model architectures and inference systems for high throughput, low latency and reliable production performance.
• Improve compute efficiency, GPU utilisation and cost per call or minute through techniques such as quantisation, batching, distillation and inference optimisation.
• Establish rigorous benchmarks for model quality, latency, throughput, reliability and cost, ensuring improvements translate into measurable customer outcomes.
• Take models from research and experimentation to production at enterprise scale, with robust monitoring, regression testing and continuous improvement. 4. Own Technical Direction and Research
• Own the technical roadmap for GreyLabs’ speech and language models, identifying high-impact research and engineering opportunities.
• Make build-versus-adapt decisions across open-weight models, fine-tuning, distillation and training from scratch, based on quality, latency, cost and strategic differentiation.
• Build robust datasets, evaluation frameworks and experimentation pipelines using representative, appropriately governed conversational data.
• Stay at the forefront of advances in speech and language modelling and translate relevant research into practical product capabilities.
• Make critical architecture and research investment decisions based on technical merit, scalability, quality and business impact.
• Establish high standards for scientific rigour, reproducibility and engineering excellence. 5. Mentor and Multiply Technical Talent
• Mentor and collaborate with research engineers, applied scientists and ML engineers.
• Lead through technical depth: review architectures, challenge assumptions, guide experiments and help solve the most difficult problems.
• Help recruit and develop a high-calibre team while remaining personally involved in critical model-building work.
• Foster a culture of experimentation, ownership, technical excellence and measurable outcomes. What We Are Looking For
• Substantial experience in AI, machine learning, speech technology or related fields, typically 13–15+ years, with a strong record of hands-on model building.
• Deep expertise in at least one of speech recognition, speech-language modelling or language model development,



with the ability to influence work across the broader speech and LLM stack.
• A proven track record of personally building, training, fine-tuning or materially improving models—not just integrating APIs or managing teams that do the work.
• Strong foundations in deep learning, model architectures, optimisation and experimentation, with hands-on proficiency in Python, PyTorch and modern ML frameworks.
• Experience taking models from research to production, including evaluation, inference optimisation, deployment and performance at scale.
• Demonstrated ability to make sound technical trade-offs across quality, latency, computational requirements and cost.
• Ability to operate independently, challenge conventional approaches and take ownership of complex technical problems.
• A track record of mentoring engineers or researchers and influencing technical direction through strong individual technical contributions. Strongly Preferred
• Experience at a frontier AI research lab, a research-intensive technology company or a startup that builds its own models.
• Experience with multilingual ASR, Indian languages, code-switching, speech processing or challenging real-world audio.
• Experience with large-scale training, LLM post-training, model distillation, inference optimisation or model-serving systems.
• An MS or PhD in Computer Science, Machine Learning, Speech Processing or a related field—or equivalent demonstrated technical depth.
• Significant technical contributions demonstrated through research publications, open-source projects, model releases or production deployments. What Success Looks Like
• Establish a transparent, outcome-oriented technical roadmap for GreyLabs’ speech and language models.
• Deliver measurable improvements in recognition accuracy, conversational quality, latency, reliability and cost per interaction.
• Take improved models from experimentation into production at enterprise scale.
• Reduce dependence on expensive proprietary model APIs where open-weight or internally adapted models can meet quality and performance requirements.
• Establish robust evaluation and research-to-production workflows that enable continuous model improvement.
• Build differentiated technical capabilities for Indian enterprise Voice AI and support conversations at massive scale. Why Join GreyLabs AI?
• Build AI that matters: Develop models with the ambition to power conversations touching billions of lives every day.
• Own a major technical charter: Shape the research, architecture and evolution of core AI capabilities.
• Stay hands-on: Solve challenging modelling and engineering problems rather than moving exclusively into people management.
• Build for real-world impact: Take ideas from research to production systems where quality, scale and cost all matter.
• Influence the future of AI: Work with talented engineers and researchers and help establish a world-class model-building capability.

📌 Distinguished Research Engineer (Bengaluru)
🏢 GreyLabs AI
📍 Bengaluru