Principal/Senior/Software Engineer (AI-Powered Advertising Agents - AgenticOS)

PubMatic; Maharashtra; On-site; INR 1,800,000 - 3,200,000; Full time

Benefits offered by this job

paternity/maternity leave

healthcare insurance

broadband reimbursement

kitchen loaded with healthy snacks and

drinks and catered lunches

Job summary

PubMatic is seeking engineers with Generative AI expertise to build and optimize AI agents using RAG, vector databases, and LLMs. You will own end-to-end design, testing, and deployment in an Agile setup.

You will work with LangGraph, FAISS, Pinecone, Weaviate, and other tools to deliver scalable AI-driven features that improve retrieval, generation, and user interactions across PubMatic products.

Qualifications

• Experience with LLMs and transformer architectures.
• Hands-on with AI agents and multi-agent systems.
• Familiarity with RAG pipelines and vector databases.

Responsibilities

• Lead design, development, and deployment of AI-driven features.
• Create scalable and secure AI architectures and design docs.
• Implement and optimize LLMs for specific use cases.
• Develop AI agents powered by RAG with external knowledge sources.
• Design and optimize vector databases and indexing algorithms.
• Craft and refine prompts for context-sensitive outputs.
• Collaborate with data scientists and engineers to ship customer-facing features.

Skills

LLMs fundamentals

Agent orchestration

Prompt engineering

Model evaluation

Education

Bachelor’s degree in engineering

Tools

LangGraph

CrewAI

AutoGen

FAISS

Pinecone

Weaviate

Milvus

Langfuse

Python

TensorFlow

PyTorch

Hugging Face Transformers

Job description

About The Role

PubMatic is looking for engineers with expertise in Generative AI and AI agent development. You will be responsible for building and optimizing advanced AI agents that leverage the latest technologies in Retrieval-Augmented Generation (RAG), vector databases, and large language models (LLMs). You will work on developing state-of-the-art solutions that enhance Generative AI capabilities and enable our platform to handle complex information retrieval, contextual generation, and adaptive interactions.

What You'll Do

• Lead the design, development, and deployment of AI-driven features. Drive end-to-end ownership—from feasibility analysis and design specifications to execution and release—while ensuring quick iterations based on customer feedback in a fast-paced Agile environment.
• Spearhead technical design meetings and produce detailed design documents that outline scalable, secure, and robust AI architectures.
• Ensure that the solutions are aligned with long-term product strategy and technical roadmaps.
• Implement and optimize LLMs for specific use cases, including fine-tuning models, deploying pre-trained models, and evaluating their performance.
• Develop AI agents powered by RAG systems, integrating external knowledge sources to improve the accuracy and relevance of generated content.
• Design, implement, and optimize vector databases (e.g., FAISS, Pinecone, Weaviate) for efficient and scalable vector search, and work on various vector indexing algorithms.
• Create sophisticated prompts and fine-tune them to improve the performance of LLMs in generating precise and contextually relevant responses.
• Utilize evaluation frameworks and metrics (e.g., Evals) to assess and improve the performance of generative models and AI systems.
• Work with data scientists, engineers, and product teams to integrate AI-driven capabilities into customer-facing products and internal tools.
• Stay up to date with the latest research and trends in LLMs, RAG, and generative AI technologies to drive innovation in the company’s offerings.
• Continuously monitor and optimize models to improve their performance, scalability, and cost efficiency.

We'd Love For You To Have

• 2 to 10 years of experience and strong understanding of LLMs and their underlying principles — transformer architecture, attention mechanisms, and hyperparameter tuning.
• Proven experience designing and building AI agents, including multi-agent orchestration, tool-use patterns, multi-step planning, and agent memory architectures (short-term and long-term).
• Hands-on experience with agentic frameworks such as LangGraph, CrewAI, or AutoGen, and familiarity with RAG pipelines that integrate external knowledge sources (documents, databases, APIs).
• In-depth knowledge of vector databases and indexing algorithms; practical experience with FAISS, Pinecone, Weaviate, or Milvus.
• Experience with agent observability, tracing, and guardrails — tools like Langfuse or equivalent — to ensure reliability, safety, and debuggability of agentic systems.
• Proficiency in prompt engineering — crafting, iterating, and optimizing complex prompts for context-sensitive, domain-specific LLM outputs.
• Familiarity with Evals and other performance evaluation tools for measuring model quality, relevance, and efficiency.
• Proficiency in Python and experience with machine learning libraries such as TensorFlow, PyTorch, and Hugging Face Transformers.
• Experience with data preprocessing, vectorization, and handling large-scale datasets.
• Ability to present complex technical ideas and results to both technical and non-technical stakeholders.

Nice-to-Have

• Experience in building AI agents using graph-based architectures, including knowledge graph embeddings and graph neural networks (GNNs).
• Experience with training small base models using custom data, including data collection, pre-processing, and fine-tuning models to specific domains or tasks.
• Familiarity with deploying AI models on cloud platforms (AWS, GCP, Azure) and containerization technologies (Docker, Kubernetes).
• Familiarity with programmatic advertising, RTB, or ad auction mechanics.
• Knowledge of MCP (Model Context Protocol) or similar tool-integration standards
• Publication or contributions to research in AI, LLMs, or related fields.

Qualification

Should have a bachelor’s degree in engineering or an equivalent degree from a well-known institute/university.

Additional Information

Return to Office

PubMatic employees throughout the globe have returned to our offices via a hybrid work schedule (3 days “in office” and 2 days “working remotely”) that is intended to maximize collaboration, innovation, and productivity among teams and across functions.

Benefits

• paternity/maternity leave
• healthcare insurance
• broadband reimbursement
• kitchen loaded with healthy snacks and drinks and catered lunches and much more!

Diversity and Inclusion

PubMatic is proud to be an equal opportunity employer; we don’t just value diversity, we promote and celebrate it. We do not discriminate on the basis of race, religion, color, national origin, gender, sexual orientation, age, marital status, veteran status, or disability status

About PubMatic

PubMatic is one of the world’s leading scaled digital advertising platforms, offering more transparent advertising solutions to publishers, media buyers, commerce companies and data owners, allowing them to harness the power and potential of the open internet to drive better business outcomes. Founded in 2006 with the vision that data-driven decisioning would be the future of digital advertising, we enable content creators to run a more profitable advertising business, which in turn allows them to invest back into the multi-screen and multi-format content that consumers demand.