Basics

Compensation: Competitive Market Salary

Location: HSR, Bangalore (On-site)

Kavana is redefining how humans engage with AI - we are the first and probably the only company building for entertainment with AI. Imagine stepping into a world where you can chat/roleplay with a gangster, an astrologer, a storyteller, a fitness coach, or even a pirate from the Arabian Sea — all powered by AI.

Who we’re looking for

We’re looking for an Applied AI Engineer who can turn large models into delightful, nearly human character experiences. You’ll be responsible for building, fine-tuning, evaluating, and self-hosting LLMs to create emotionally engaging conversational experiences.

This role combines AI research with engineering execution - from training models that remember personalities to optimising them for latency, emotion, and engagement metrics.

You’ll work directly with the founders to bring AI entertainment to life.

About the Role

Responsibilities:

• Fine-tune and optimise LLMs (instruction-tuned or domain-specific) for conversation quality, memory consistency, and character realism.
• Build and optimise voice and text-based AI models & components for natural, engaging & conversational storytelling.
• Build and manage self-hosted inference pipelines for large language and speech models.
• Define and own core engagement metrics (e.g., message depth, retention, engagement time, response quality, etc.) and improve them through experimentation.
• Develop evaluation frameworks for empathy, memory, contextual understanding, human-likeness, tone, and narrative depth.
• Collaborate with the backend and product teams to deploy scalable, performant conversational systems for all our products.

Requirements:

• 2–4 years of hands-on experience building and deploying generative AI models in production.
• Strong understanding of generative AI modelling fundamentals: embeddings, fine-tuning, evaluation loops, and inference optimisation.