About this role
Lead AI Engineer for Merkle, based in Pune, developing generative AI solutions for customer engagement. The role partners with engineering, data science, customer experience, and product teams to build intelligent agents, multimodal experiences, personalization systems, and knowledge-grounded applications for global brands.
Responsibilities
- Build conversational, non-conversational, multimodal, and agentic applications using LLMs and frameworks such as LangChain, LangGraph, LlamaIndex, or AutoGen. Design workflows with reasoning, planning, tool use, memory, grounding, and external integrations. Develop knowledge-graph-assisted systems with entity extraction, linking, and augmented retrieval; apply evaluation and guardrails to improve safety and control hallucinations.
- Build and optimize transformer-based and multimodal models using frameworks such as PyTorch and TensorFlow. Implement fine-tuning, RLHF/RLAIF alignment, LoRA/QLoRA, pruning, distillation, and evaluation. Develop hybrid dense–sparse retrieval, ranking, and relevance systems, alongside predictive models and end-to-end ML pipelines.
- Deliver scalable Python APIs and microservices using FastAPI or Flask and Docker. Deploy and monitor AI systems on AWS, Azure, or GCP, balancing cost, latency, and reliability. Work with MLOps teams on CI/CD, versioning, monitoring, and automated evaluation, and use Spark, Hadoop, and MongoDB for large-scale data processing.
- Document experiments, models, evaluation frameworks, and deployments; mentor junior engineers and contribute reusable components and R&D work.
Requirements
Minimum 5–6 years of hands-on software development experience, including building and deploying ML models into production, plus 2+ years with deep learning, GenAI, or transformer-based architectures. Candidates must have built GenAI applications beyond basic RAG, such as agents, multimodal systems, or custom LLM fine-tuning, and integrated AI in enterprise environments. Required capabilities include advanced Python, SQL, API development, data engineering, backend frameworks, ML and retrieval techniques, cloud deployment, data pipelines, and foundations in linear algebra, probability, and statistics. Clear communication with non-technical stakeholders, collaboration, and problem-solving are expected.
Arrangement: Full-time, permanent, hybrid in Pune, with at least two days working from the office.