AI Engineer

We're looking for an exceptional AI/NLP Engineer to build the intelligence behind our enterprise AI Agents. You'll work closely with the CTO and AI Research team to develop, fine-tune, and deploy state-of-the-art AI models powering multilingual, context-aware conversations at scale. If you're passionate about applied AI, LLMs, speech technologies, and solving production AI challenges, we'd love to hear from you.

Responsibilities:

• Design and build high-quality datasets for training and evaluation, including synthetic data generation, data annotation pipelines, and quality assurance
• Conduct applied research to improve agent performance in multilingual settings, intent classification, sentiment analysis, and context understanding
• Optimise and fine-tune TTS models for naturalness, emotion control, and low-latency performance across multiple languages.
• Develop and improve ASR/STT systems for accurate transcription in noisy environments, handling code-mixed speech and domain-specific vocabulary.
• Fine-tune and adapt large language models (LLMs) for industry-specific use cases, including collections, customer service, and sales conversations.
• Develop sophisticated prompt engineering strategies and few-shot learning approaches to optimise model performance.
• Design comprehensive evaluation frameworks, including automated testing, human-in-the-loop evaluation, and A/B testing methodologies.
• Build and optimise Retrieval-Augmented Generation (RAG) pipelines for grounding AI responses in regulatory documents and financial knowledge bases.
• Develop and optimise models for Indian languages (Hindi, Tamil, Telugu, Bengali, etc. ) with a focus on code-mixing and transliteration.
• Implement model monitoring, drift detection, and continuous improvement pipelines for production AI systems.

Requirements:

• 3+ years of experience in AI/NLP with production deployments.
• Expert-level experience fine-tuning language models using: Parameter-efficient fine-tuning (LoRA, QLoRA, Adapter layers), Instruction tuning and RLHF concepts, Domain adaptation techniques.
• Fine-tuning and optimising neural TTS models (Tacotron, FastSpeech, VITS, or similar).
• Working with ASR architectures (Whisper, Wav2Vec2 Conformer, or similar).
• Proven experience building high-quality training datasets, including Synthetic data generation techniques and Data augmentation approaches.
• Understanding of model serving, API integration, and latency optimisation.

Strong proficiency in:

• PyTorch with hands-on model training experience.
• Solid understanding of Transformer architectures (attention mechanisms, positional encoding).
• Pre-training and transfer learning strategies.
• Tokenisation, embeddings, and language representation.