Staff Engineer, AI Engineer

Job Description

Bully Pulpit International is an outcomes agency made up of strategists, data scientists, and artists. We come from politics, brands, and government and we communicate without barriers and measure what matters. We have more than 350 team members in six countries and thirteen offices — Berlin, Brussels, Chicago, D.C., London, Los Angeles, New York, Oslo, San Francisco, and Zürich. We apply our unique expertise across strategic communications, public affairs, research, and digital marketing to create sophisticated and customized strategies that make change possible for both American and European clients.

• Starting Salary Range: $140,000 - $150,000
• Location:
  Expectation to work from DC, New York, Chicago, SF or LA offices at least 3 days a week

Our Labs team builds new tools and capabilities across BPI's technology stack. Our team's focus will always be on building software that solves real problems for our teams and clients. We're also entering a phase where we want to make it possible for teams beyond product engineering—analysts, consultants, operations—to build LLM-powered prototypes that also solve real problems. But we also want to do this responsibly: with the right guardrails, evaluation practices, and pathways to move the best ideas into production safely.

We're looking for a Staff Engineer to help us figure out what that looks like. You'll build new tools and products directly as part of the Labs team, while also defining the solutions (process, standards, tooling, and infrastructure) that enable teams across BPI—many already prototyping with tools like Replit and LLM-powered coding assistants—to move their best ideas safely into production. This is a role for someone who is both a builder and a thinker—someone who gets to build great software directly using LLM agents, but who also asks hard questions about reliability, security, and maintainability before shipping something to users.

• Write production-quality code across the full stack, working across the entire software lifecycle from shaping features with stakeholders to deployment and optimization
• Build new tools and products from scratch as part of the Labs team—prototype-hardening is a major part of this role, but you'll also own and ship your own projects end-to-end
• Define the overall approach for how prototypes built by teams across BPI—often in tools like Replit—get evaluated, hardened, and moved into production (or don’t)
• Build and own the solutions that make this cycle repeatable: the mix of process, standards, tooling, and technical infrastructure that lets this work safely at scale
• Establish evaluation, testing, and review frameworks for LLM-powered applications—including knowing when an AI-driven approach isn’t the right one
• Set technical direction and mentor other engineers and non-engineering builders as they work with LLM tools, helping them build prototypes that are easier to harden later
• Work directly with stakeholders and end users to consult on their prototypes, shape what’s worth investing in, and translate the best ideas into reliable, production-grade software

• 7+ years of experience in software development, building, launching, and maintaining full-stack applications
• A strong foundation in software engineering: you write clean, well-tested code and care about architecture, security, and maintainability
• Strong experience building production systems using Python/Django and React
• Strong experience working with cloud infrastructure (AWS, GCP, or Azure), containers (Docker), and CI/CD pipelines
• Experience with data tools like Snowflake/dbt is a significant plus
• Plenty of experience with a wide variety of SDLCs, and a proven ability to influence process in pragmatic and non-dogmatic ways
• Hands-on experience building with LLM APIs, agent frameworks, retrieval-augmented (RAG), and prompt engineering—you’ve shipped LLM-powered features or products, not just experimented
• Product-focused thinking—you’re comfortable in a room with stakeholders, shaping what to build, not just how to build it
• A healthy skepticism about AI hype, grounded in practical concerns: you…