Senior AI Engineer

Why This Role Exists

PubX builds next-generation Agentic Advertising infrastructure. Our Agentic AI makes real-time, revenue-critical decisions for digital publishers and advertisers. Our Bid Intelligence uses machine learning to optimize every programmatic ad auction individually, generating measurable revenue uplift for publishers. We've priced over 1 trillion programmatic auctions, we’re currently ranked #5 globally in Prebid Analytics Adapter Rankings, and growing.

The problem we’re solving

Digital publishers and advertisers leave significant revenue on the table because ad sales are still largely manual, static, or rule-based. The reality is every campaign is unique, and most ad campaign management systems lack the intelligence to evaluate context, demand, and deal potential in real time.

As a founding member of AgenticAdvertising.org, we're building the next generation of autonomous advertising infrastructure.

Who You Are

• A person with a pragmatic engineering mindset: you know when to use an agent framework, when a simpler automation workflow is better, and how to balance model capability, reliability, latency, cost, and user experience.
• Communicate technical ideas well in writing and conversation to both technical and non-technical audiences.
• Have taste for clean, well-tested code with thoughtful abstractions that’s easy to extend and operate with agents.
• Learn quickly when things are unfamiliar by prototyping, then hardening and documenting what you ship.

What You’ll Work On

• Design, build, and maintain backend services and APIs across REST, GraphQL, and gRPC that power agentic AI features and core product workflows.
• Build production-grade AI agent systems using modern frameworks and patterns such as LangChain, LangGraph, tool/function calling, structured outputs, multi-step planning, stateful agent sessions, human-in-the-loop workflows, and agent memory.
• Design and integrate agent interoperability patterns, including Model Context Protocol (MCP) for tool and data access, and Agent-to-Agent (A2A) communication patterns for multi-agent collaboration.
• Develop web applications that integrate tightly with ML and LLM-driven systems, including, stateful conversations, retries, fallbacks, and explainability surfaces.
• Build and operate event-driven components using Kafka and message queues such as SQS or RabbitMQ, designing asynchronous workflows for long-running agent jobs, background reasoning tasks, tool execution, and orchestration pipelines.
• Implement observability, evaluation, and debugging workflows for AI systems using tools such as Langfuse, LangSmith, OpenTelemetry, structured logs, metrics, tracing, agent/tool traces, dashboards, alerting, and post-incident reviews.
• Deploy and operate services on AWS, including ECS, EKS, Lambda, API Gateway, ALB, RDS, DynamoDB, S3, and related services, using infrastructure-as-code and secure-by-default engineering patterns.
• Continuously improve reliability, performance, cost efficiency, developer experience, and production-readiness across both traditional backend systems and AI-native workflows.

What We’re Looking For

We’re looking for an experienced engineer who has built production systems and enjoys solving practical problems with AI. You should be comfortable working across backend engineering, cloud infrastructure, and LLM-powered product experiences.

You’ve likely worked with:

• Strong backend engineering fundamentals, including API-first design, data modeling, distributed systems, performance, security, authentication, authorization, and production operations.
• Hands-on experience integrating LLMs or ML components into real product systems, including streaming responses, async jobs, state management, caching, retrieval, grounding, evaluations, and monitoring.
• Practical experience with agentic AI frameworks and patterns, such as LangChain, LangGraph, LlamaIndex, OpenAI/Anthropic APIs, tool calling, structured outputs, agent routing, multi-agent workflows, and human-in-the-loop systems.
• Experience with LLM observability and evaluation workflows, ideally using tools such as Langfuse, LangSmith, OpenTelemetry, custom evaluation harnesses, prompt/version management, datasets, regression testing, and LLM-as-judge approaches.
• Strong hands-on experience with AI-assisted development workflows in a real codebase, not just demos, including using coding agents to navigate large repositories, propose multi-file changes, write and update tests, review diffs, and iterate safely.
• Experience working with distributed systems with event-driven patterns, including Kafka, SQS, RabbitMQ, pub/sub architectures, background workers, retries, dead-letter queues, idempotency, and workflow orchestration.
• Hands-on AWS experience with production operations, monitoring, incident response, infrastructure-as-code, cost awareness, and secure deployment practices.

Nice to Have

• Experience with MCP servers or clients, A2A-style agent communication, tool registries, agent authentication, permissions, audit logging, and secure tool execution.
• Experience with retrieval-augmented generation systems, vector databases, embedding pipelines, reranking, hybrid search, citation generation, and grounding quality evaluation.
• Experience designing safeguards for LLM systems, including prompt-injection mitigation, output validation, PII handling, policy enforcement, sandboxing, rate limiting, and abuse monitoring.
• Experience with modern AI developer tooling such as Cursor, Claude Code, GitHub Copilot, OpenAI Codex, or similar coding-agent workflows in production engineering teams.

Bonus (not required):

• Experience with AdTech or other high volume real-time systems

Who This Role Will Suit

• This role suits engineers who like a mix of autonomy and collaboration, and who are comfortable working in an environment that’s still evolving.
• We’re a distributed team with a growing engineering presence in India, so comfort with async collaboration and clear written communication is important.
• We use agentic coding tools heavily (e.g. Cursor and Claude Code) to plan, scaffold, refactor, and debug production code, while maintaining strong engineering judgment and ownership of outcomes.

Company Benefits

• Competitive salary with meaningful equity
• Fully remote, async-friendly working
• Supportive, low-ego engineering culture
• Budget for learning and professional development

Interview Process

Our process is designed to be practical and respectful.

• CV & Profile Review – Relevant experience and background
• Initial Chat (30 mins) – Motivation and role fit
• Architecture Interview (60 mins) – Architecture, design choices, and real scenarios
• Agentic Coding Exercise (60 mins) – A small task related to the role, executed by agents

If you’re interested in building and shaping real systems in a growing product company, at the forefront of AdTech innovation, we’d love to hear from you.

We will process your personal data in accordance with our Recruitment Privacy Notice: https://pubx.ai/privacy/recruitment/