Description :

Role : AI/ML Engineer (LLMs, RAG & Agent Systems)

Location : Bangalore, India (On-site)

Type : Full-time

About the Role :

As an AI/ML Engineer, youll be part of a small, fast-moving team focused on developing LLM-powered agentic systems that drive our next generation of AI products.

Youll work on designing, implementing, and optimizing pipelines involving retrieval-augmented generation (RAG), multi-agent coordination, and tool-using AI systems.

Responsibilities :

- Design and implement components for LLM-based systems (retrievers, planners, memory, evaluators).

- Build and maintain RAG pipelines using vector databases and embedding models.

- Experiment with reasoning frameworks like ReAct, Tree of Thought, and Reflexion.

- Collaborate with backend and infra teams to deploy and optimize agentic applications.

- Research and experiment with open-source LLM frameworks to identify best-fit architectures.

- Contribute to internal tools for evaluation, benchmarking, and scaling AI agents.

Required Skills :

- Strong foundation in ML/DL theory and implementation (PyTorch preferred).

- Understanding of transformer architectures, embeddings, and LLM mechanics.

- Practical exposure to prompt engineering, tool calling, and structured output design.

- Experience in Python, Git/GitHub, and data processing pipelines.

- Familiarity with RAG systems, vector databases, and API-based model inference.

- Ability to write clean, modular, and reproducible code.

Preferred Skills :

- Experience with LangChain, LangGraph, Autogen, or CrewAI.

- Hands-on with Hugging Face ecosystem (transformers, datasets, etc.).

- Working knowledge of Redis, PostgreSQL, or MongoDB.

- Experience with Docker and deployment workflows.

- Familiarity with OpenAI, Anthropic, vLLM, or Ollama inference APIs.

- Exposure to MLOps concepts like CI/CD, model versioning, or cloud (AWS/GCP/Azure).

What We Value :

- Deep understanding of core principles over surface-level familiarity with tools.

- Ability to think like a researcher and execute like an engineer.

- Collaborative mindset, building together, learning together.