AI/ML Engineer Job ID: 397504

Kuku FM, Inc.; Mumbai; On-site; INR 1,200,000 - 2,400,000; Full time

Job summary

Kuku FM, Inc. is seeking an AI/ML Engineer to help build its next-generation AI-powered content creation platform in Mumbai.

You will focus on LLMs, multimodal AI, content understanding, and AI-generated microdramas, handling large-scale transcripts and production pipelines. You'll collaborate with senior engineers and product teams to deploy scalable ML systems, optimize models, and design end-to-end workflows for screenplay generation, video prompts, and semantic retrieval across the content

Qualifications

• 2–4 years of relevant ML experience with deployment.
• Hands-on experience with LLM APIs and GenAI applications.
• Experience with PyTorch or TensorFlow for model training.

Responsibilities

• Build LLM-powered systems for screenplay and dialogue generation.
• Curate instruction-tuning data from large-scale transcripts.
• Develop ML pipelines to extract characters, scenes, and arcs.
• Transform ideas into episodes, scenes, and screenplays.
• Create workflows for AI video generation prompts.
• Design LangGraph/LangChain workflows for planning and checks.
• Build semantic retrieval for characters, stories, and content.
• Experiment with LLMs, VLMs, and embeddings for UX, latency, cost.
• End-to-end ML execution: data pipelines, inference, monitoring.
• Collaborate with senior ML engineers, product teams, backend engineers.

Skills

Python programming

ML engineering

GenAI

LLM APIs

PyTorch/TensorFlow

NLP

Production ML

Data processing

Cloud platforms

Docker/Kubernetes

Education

Bachelor's or Master's in CS/ML

Tools

Hugging Face Transformers

LangChain

LangGraph

Seedance

PyTorch

Docker

Kubernetes

Job description

About the job

As an AI/ML Engineer, you will work on building Kuku's next-generation AI-powered content creation platform, with a focus on LLMs, multimodal AI, content understanding, and AI-generated microdramas. You will work with large-scale microdrama transcripts and build ML systems that can understand stories, characters, scenes, and narrative structures, and use this knowledge to generate and optimize new content.

This role provides an opportunity to work across Generative AI, LLMs, instruction-tuning data, agentic workflows, screenplay generation, and AI video generation, while building production-grade ML systems used by Kuku's content ecosystem.

Responsibilities

• LLM & GenAI: Build LLM-powered systems for screenplay generation, dialogue generation, story continuation, summarization, rewriting, critique, and content transformation.
• Instruction-Tuning Data: Build and curate high-quality instruction-tuning and evaluation datasets from Kuku's large-scale microdrama transcript corpus.
• Content Understanding: Develop ML/LLM pipelines to extract characters, relationships, scenes, story arcs, dramatic beats, emotions, conflicts, twists, and other structured information from content.
• AI Screenwriting: Build systems that transform story ideas and structured story information into episodes, scenes, dialogue, and production-ready screenplays.
• AI Video Generation: Develop workflows to convert screenplay and scene information into structured prompts and instructions for AI video generation models such as Seedance.
• Agentic Systems: Build multi-step LLM workflows using frameworks such as LangGraph or LangChain for story planning, screenplay generation, quality checking, continuity validation, and video prompt generation.
• Embeddings & Retrieval: Build semantic retrieval systems for characters, stories, scenes, dialogue, and other content using embedding models and vector databases.
• Model Experimentation: Experiment with different LLMs, VLMs, embedding models, fine-tuning techniques, prompting strategies, and retrieval approaches to improve quality, latency, and cost.
• Evaluation: Develop automated and human-in-the-loop evaluation pipelines to measure story quality, narrative consistency, character consistency, and AI-generated content quality.
• ML Execution: Build and productionise ML systems end-to-end, including data pipelines, model inference, experimentation, monitoring, and optimization.
• Collaboration: Work closely with senior ML engineers, content teams, product managers, and backend engineers to translate creative and product problems into scalable ML solutions.

Preferred Qualifications

• Education & Experience: Bachelor's or Master's in Computer Science, Machine Learning, Statistics, or a related engineering field, with 2-4 years of relevant experience and a proven ability to build and deploy ML systems
• LLM & GenAI: Hands-on experience with LLM APIs, open-source LLMs, prompt engineering, RAG, embeddings, or generative AI applications.
• Model Training: Experience training or fine-tuning ML/deep learning models using frameworks such as PyTorch or TensorFlow. Familiarity with Hugging Face Transformers and LoRA/PEFT is a plus.
• Production Experience: Experience building and deploying production ML models or GenAI applications, including data pipelines, APIs, monitoring, and cloud infrastructure.
• NLP: Strong understanding of transformers, embeddings, semantic search, text classification, sequence modeling, or related NLP techniques.
• Agentic Systems: Experience with LLM orchestration frameworks such as LangGraph, LangChain, or equivalent technologies is desirable.
• Multimodal AI: Experience with VLMs, speech/audio models, image generation, or video generation is a strong plus.
• ML Fundamentals: Strong understanding of machine learning, deep learning, experimentation, and model evaluation.
• Engineering: Strong Python programming and software engineering skills with the ability to build scalable and maintainable ML systems.
• Data & Infrastructure: Experience with SQL, data processing, vector databases, cloud platforms, Docker, Kubernetes, or similar technologies is desirable.
• Research Awareness: Stay up-to-date with advancements in LLMs, multimodal AI, generative AI, and applied machine learning.