About this role
Summary: Lead the expansion and modernization of Firn, the company's unified data and AI platform. Design and build GTM intelligence engines, agents, and tools while establishing production-grade architecture, CI/CD, and evaluation infrastructure.
### Boston/Hybrid or Massachusetts/Remote
About Snowplow
Snowplow is the real-time customer context layer that collects, validates, enriches, and delivers behavioral data for advanced analytics, ML, and AI agent decisioning. Snowplow's event tracking is leveraged across 2M+ websites and applications globally, processing over one trillion events per month.
More than 250 companies, including Experian, AutoTrader, Strava, Condé Nast, and HelloFresh, rely on Snowplow to build a well-governed, first-party data foundation that powers their customer-facing AI agents, in-session personalization and recommendations, and real-time analytics.
This Opportunity
Snowplow is rebuilding how the company runs its own work, and the Applied AI team owns Firn, the single data and AI platform the whole company runs on. Firn already gives people and agents the same governed tools over our Snowflake warehouse through an MCP gateway, routes every model call through one LLM gateway with tracing, runs agentic workflows on a factory framework with a ledger that records every run, and holds the skills catalog the company draws on. Most of the company uses it every day. The platform is at its beginning, and the next stage is to expand it until it holds the company's commercial intelligence and lets agents act on it safely.
You are the principal engineer who leads that expansion. The first engagement is Go-to-Market, full time for about six months. Our buyers research with AI agents now, and reaching them means using our own data and AI to decide which accounts and people to engage, with what, and learning from every experiment. The GTM teams have built a lot: daily account and engagement scoring, signal feeds from Apollo, Vector, Kettlepoint and reo.dev, an always-on outbound engine, a paid-media loop, a curated GTM knowledge base, and the integrations between them. Much of it was built by operational teams to get results fast, and it did. You design how those services come onto Firn, unify and modernize them, build the agents and tools and learning loops on top, and make the whole thing production grade.
This is a principal role. You report to the Senior Director of Applied AI, who still ships code, and work as a peer. Marketing, Sales, Account Management and Partnerships are the platform's customers and set their own priorities. You decide how it gets built and whether it is sound. Once the GTM services run without you, they stay yours as their owner and most of your time goes to what Snowplow needs next.
What you will own
- The GTM intelligence engine on Firn. The data and AI that decide who we engage and with what, and that learn from the outcome of every action. You design it with our data engineers as a set of platform services with several internal customers, and you bring the GTM services that exist today onto the same foundation.
- Agents and tools. The agents that run GTM work and the tools people and other agents call, built on the factory framework with holdback gates, review steps and evaluations as the standard shape, so every agent on the platform can be trusted and inspected the same way.
- Platform architecture. The MCP gateway, LLM gateway, tracing, factory framework and ledger, and how they grow to hold more of the company's work. You decide what is built, what is bought, and what is retired.
- Production grade. Version control, CI, alerting and recovery, cost visibility, documentation, and handover for every service on the platform, including the ones that arrive from operational teams. You set the bar and you make it visible to the teams that depend on it.
- Logic as code. Rules that today live inside individual tools move into versioned code on the platform, where agents can read them and engineers can test them, and the tools become endpoints the platform drives.
- Learning loops. Every action the platform takes is recorded and tied to its outcome, so people and agents can see what worked and the next action is better. Outcomes usually land days after the run, so the ledger carries a nightly true-up against CRM and billing events rather than trusting what the agent assumed at the time. Metric definitions sit with our data engineers; the instrumentation that applies them is yours.
- Eval and tracing infrastructure. The harness the embedded and forward deployed engineers use to prove their agents work. You own the tooling; they own the checks.
- The technical bar and the roadmap. You set the engineering bar for the Applied AI team and own the platform roadmap, and you keep the long view of the platform at all times.
We would love to hear from you if you have
- Years of production software behind you. You have designed, shipped and run systems other teams depend on, in Python or TypeScript, with SQL, CI/CD and cloud infrastructure as everyday tools. This is a principal role and the bar is set there.
- Built LLM systems that real work runs on. Agents with tools, MCP servers, gateways, tracing and evaluation, and the failure modes that come with them. You can say what broke in production and what you changed.
- Platform instincts. You have built systems other people depend on, whether internal teams or customers, through interfaces you designed, and you know the difference between a platform and a pile of integrations.
- Data platform fluency. Snowflake or an equivalent warehouse, dbt, data contracts, and the pipelines that move warehouse data into SaaS tools and back.
- Operational discipline. Alerting, recovery, runbooks and cost tracking are things you add before anyone asks, and you have run systems where that discipline mattered.
- Judgment about where an agent belongs. You can tell which decisions should be deterministic code, which an agent can make, and which should stay with a person, and you can explain the reasoning to a non-engineer.
- An AI-native way of working you can show us. A workflow, agent or tool you built and still use. Expect to demonstrate it during the interview process.
- Comfort with internal customers who are not engineers. You can sit with a growth director, a seller or a marketer, take a requirement, and hold the line on how it gets built. No sales or marketing background is needed.
- Technical direction without a title. You have set direction for a platform that other engineers followed because the design was right, not because they reported to you.
How we hire for this role
The interview process includes a working session where you walk us through how you work with AI on a real task, using your own setup, and a design conversation on a system we run today. We work this way ourselves, and you are welcome to use AI when applying.
What you get in return for being awesome
- Meaningful equity stake
- Competitive Base Salary
- Flexible working
- Generous PTO
- Medical, Dental & Vision Insurance
- 401k
- MacBook and home office equipment allowance
- Enhanced Family leave