We are expanding our agentic AI capability and are looking for an AI Engineer to join the team. You will work alongside senior engineers to build and maintain AI systems — contributing to agentic pipelines, retrieval infrastructure, and the integrations that tie these systems together.

This is a hands-on implementation role with real ownership of components. You will grow your skills in a rapid-moving AI practice, working on production systems that directly affect client outcomes.

What This Involves:
- Build and maintain agentic pipelines and workflows under the guidance of senior engineers: tool use, orchestration, and multi-step reasoning.
- Implement and tune RAG pipelines — including embedding, chunking strategies, vector retrieval, and retrieval evaluation.
- Contribute to memory and context layer components: integrating vector databases, supporting knowledge graph pipelines, and helping maintain state management across agentic systems.
- Write clean, well-tested Python code and participate in code reviews.




- Debug and improve existing AI systems based on evaluation results and production feedback.
- Collaborate with data engineers and domain experts to integrate AI components with upstream data sources and downstream applications.
- Document implementations clearly and contribute to shared internal tooling.

Requirements:
- 2–4 years of software or ML engineering experience, with at least 1 year working with LLMs or AI systems in a professional setting.
- Working knowledge of LLM APIs (OpenAI, Anthropic, or similar) and at least one agentic or RAG framework (LangChain, LlamaIndex, or equivalent).
- Solid Python skills and comfort with software engineering basics: version control, testing, REST APIs.
- Familiarity with vector databases or embedding-based search.
- Curiosity about agentic AI — you follow developments in the space and are eager to apply new techniques.
- Excellent communication and collaboration skills — comfortable working across cross-function

📌 AI Engineer (Pune)
🏢 Lynx Analytics
📍 Pune