Description
Employee #1 · Rijeka, Croatia (on-site) · Full-time · Market-rate cash + founder-level equity
About ForgeForge builds independent forecast forensics for B2B revenue teams.
Every quarter, a revenue leader commits a number to their board.
That number drives hiring, spending, and valuation, and almost nobody independently verifies its credibility.
Forge does.
We measure how accurate a company's forecast is, and we decompose the reasons it misses.
We express this as a single forecast-accuracy score, delivered first as a diagnostic, then as an ongoing quarterly governance subscription.
We integrate with HubSpot, Salesforce, and other CRMs.
We read pipeline history, not replace the CRM.
We're bootstrapped, revenue-funded, and building deliberately toward a large business with a very small team.
You are the first engineer.
The roleYou are Forge's first engineer and the person the rest of the stack gets built on.
For the first several months, you're the only engineer; you make the architecture calls, own the pipeline from the CRM API to the generated report, and ship.
When engineers two and three join, they build on what you decided.
This is a first-boots-on-the-ground role, on-site in Rijeka.
Not spec-and-execute, translate customer reality into data models and shipped software, fast.
The founder is a former CRO; you'll get the business context and the "why" directly, and you'll turn it into working systems without a 20-page brief in between.
If you want a role where you build the thing that makes a manual process obsolete, and you want to do it with AI in your toolchain, not against it, this is that role.
Your first 90 daysWeeks 1–3: Read-only ingestion from HubSpot (Salesforce next).
OAuth, pagination, rate limits, and the fact that every CRM is configured differently.
Get real opportunity and deal-history data flowing.Weeks 4–8: Deal-snapshot capture, the core of everything.
We measure how deals move over time, so you capture the point-in-time state, not just the current state.
Then the first cut of the scoring engine: compute the forecast-accuracy score and decompose it into its root-cause drivers.Weeks 9–12: Report generation.
Take the analysis and produce a governance report credible enough for a CRO to present to their board.
This is the artifact that replaces work currently done by hand (or not at all), get it shipping.
What you'll build (near-term, then trajectory)Now: the engine that lets us stop delivering reports by hand:
CRM data ingestion (HubSpot, then Salesforce), read-only, resilient to messy real-world configsDeal-snapshot / time-series capture across the pipelineThe forecast-accuracy scoring and root-cause decompositionAutomated, credible report generation (PDF or web)
Next: as customer count grows:
Customer dashboardsA CRM abstraction layer so new integrations are cheapMulti-tenant isolation, encryption, audit logging, SOC 2-readinessOperational tooling so 10 → 100 customers doesn't mean 10x manual work
You'll help decide the sequence.
But near-term velocity beats near-term polish.
We ship the engine, then harden it.
Must haveStrong backend Python: APIs (FastAPI/Flask), data pipelines, analytical logicPostgreSQL: schema design for time-series data, query optimization; you can model millions of deal snapshots and keep queries fastREST API integration: third-party APIs with real quirks (Salesforce/HubSpot rate limits, bulk behavior, field-level security, custom objects)Enough front-end to ship data-heavy UI: React/Next.js or a build tool like Lovable; clean, fast, chart-driven.
You don't need to be a front-end specialistYou build with AI as leverage: Claude Code or equivalent in your daily loop.
We measure output, not keystrokes.
One engineer who uses AI well should out-ship three who don'tYou ship, then improve.
You know the difference between "good enough now" and "this haunts us in six months," and you make that trade-off correctly and repeatedly
Nice to haveSalesforce and/or HubSpot API integration experience (saves months)RevOps, sales-tech, or analytics-product background: you already know CRM data is messy and whyEarly-stage startup experience (seed–Series A): no PM, no QA, no spec, just momentumData-pipeline / ETL tooling, cloud infra (AWS/GCP), OAuth 2.0, and multi-tenant security
Who you are5–8+ years full-stack/backend, with at least one product built from early stage and architecture decisions that aged wellYou operate without cover.
No CTO, no tech lead above you.
You decide, you own it, you're comfortable with thatYou translate business into systems.
"Show me forecast accuracy by region vs.
last quarter" becomes a schema change, an endpoint, and a component, no detailed spec requiredYou communicate clearly and flag risk early instead of burying itLanguages: Fluent in English, Croatian native, German a plus but not a requirement
CompensationMarket-rate cash salary €4,000–5,000 gross/month.
We pay in cash, not promises.
This is not an equity-for-salary trade.
The band reflects demonstrated senior delivery capabilities.
We calibrate the offer to evidence, so treat it as earned, not a floorFounder-level equity via a formal plan, real upside in a category we're defining, structured properlyEmployee #1 Status has a direct influence on every technical and product decisionA defined product, not a search for one, a validated diagnostic, a live content strategy, a clear plan.
You build the foundation for something customers are already asking for
How to applyEmail join@getforge.ai, subject line "Forge - Lead Engineer", with:
One paragraph on why this problem space fits your experienceCV or LinkedInA link to something you've built: repo, shipped product, or a write-up of an architecture decision you made.
I care how you think, not credentials.
Applications without a link don't move forwardA short answer to this (4–6 sentences): Our data source, a CRM, only exposes the current state of each deal.
We need to measure how deals change over time.
How would you capture and store point-in-time snapshots so we can reconstruct pipeline history and query it efficiently at scale?Your salary expectation and confirmation that on-site work in Rijeka works for you
One bar, stated plainly: if you can't point to an architecture decision you made that a team lived with for years — and explain why it held up — this isn't the role.
That's the level we're hiring at.
No cover letter.
No algorithm puzzles.
If your snapshot answer and your work are strong, we'll talk through how you'd build Forge's foundation and a past project you're proud of.
Forge evaluates candidates on skills, experience, and potential, nothing else.
FORGE · ForgeLab d.o.o (incorporation in process)· Rijeka, Croatia (engineering) · Zürich, Switzerland (HQ)