- Salary
- ₹5 - 16 LPA
- Experience
- 4+ yrs
- Location
- Pune, Mumbai, Gurugram, Bengaluru, Chennai, Hyderabad
- Job type
- Full-time
Required skills
- Generative AI
- Azure OpenAI
- Python
Nice to have
- Creative AI Engineer
- Generative AI Engineer
- LLM Engineer
About the role
Key Responsibilities
- Multi-Modal AI Pipeline: Build and optimize AI analysis and processing for static images and media assets, video and motion assets, and documents using third-party AI services, APIs, and multi-modal foundation models from cloud hyper-scalers.
- LLM Integration & Orchestration: Implement, manage and support AI provider abstraction layer on top of all major cloud AI hyper-scalers, including Google Gemini, and Azure OpenAI; with design for intelligent routing based on task complexity and cost
- Prompt Engineering: Develop and optimize channel-specific prompts for advertising and creative analysis and effectiveness (Facebook vs. Instagram vs. YouTube), encode domain expertise within system prompts, optimize for structured JSON output from LLM models
- RAG Implementation: Build and maintain retrieval-augmented generation workflows with pgvector, implement hybrid search capabilities (full-text + vector similarity), build and optimize embeddings for media assets.
- Custom AI Development: Enhance our proprietary optimization model, develop algorithms combining metrics from third-parties into dynamic strength scores.
- AI Agent Development: Build and manage conversational AI agents with tool-calling capabilities, implement state management for multi-turn dialogues, design autonomous decision-making for creative workflows
- AI-powered Asset Optimization: Manage and evolve our solution for AI-enabled Creative Asset optimization, leveraging creative performance analysis to automate optimization of assets, and drive automated incremental edits/improvements to creative assets
- Performance Optimization: Reduce inference latency, implement intelligent caching for embeddings, optimize token usage across providers, design cost-effective model selection strategies
Required Skills
- AI/Gen-AI Foundations: 4+ years working with LLMs (Gemini models, GPT, Claude), deep understanding of embeddings and vector search
- Multi-Modal AI: Proven experience processing creative assets, images and video with AI models, computer vision fundamentals, creative generation using multi modal capabilities, and prompt engineering patterns for both media understanding and for media generation
- Scaled AI Deployment: Experience deploying AI in production environments, handling async processing, implementing retry logic and error handling for AI APIs
- Python + AI Frameworks: Strong Python skills, experience with Cloud AI SDKs including Google Gemini and Vertex, Azure OpenAI, and others, familiarity with AI orchestration frameworks
- RAG Systems: Hands-on experience building retrieval-augmented generation solutions, using vector databases, semantic search implementation
- Creative AI Workflows: Hands-on experience building, managing and evolving creative workflows using AI models. This could include creative asset adaptation, automation, creative analysis, or creative production using AI.
Highly Valued
- Background in creative AI, creative technology, media or advertising technology
- Prior experience with other creative AI tools and frameworks
- Advanced Knowledge of prompt optimization techniques and structured output generation
- Experience with fine-tuning foundation models
- Familiarity with AI safety practices (content filtering, audit logging, watermarking)
- Understanding of advertising and media metrics (engagement, recall, attention, memorability)
AI Stack You'll Work With
- AI Models : Proprietary AI APIs (creative analysis), Vision Models, Multi-Modal AI Models, LLMs, Google Gemini, Azure OpenAI, Anthropic Claude
- Internal AI: Proprietary optimization models, custom dynamic strength algorithms, channel-specific prompt libraries
- Infrastructure: PostgreSQL + pgvector (HNSW indexes), Async processing pipelines, AI observability frameworks, Model Context Protocol (MCP) for tool integration
- Frameworks: Proprietary Django based Python framework for AI Workflows and AI orchestration
What You'll Build
- AI pipeline analysing creative assets for cognitive demand, engagement, memorability, attention effectiveness
- Generative AI system producing optimized creative variants from recommendations
- Vector similarity engine finding conceptually similar creatives across millions of assets
- Multi-modal agents reasoning across images, video, text to provide actionable insights
- Cost optimization layer routing requests to appropriate AI models
- Compliance system validating creative against advertising standards (planned)
This page is fully interactive when JavaScript is enabled. Please enable JavaScript to apply or browse related roles.