Job Description

Full-Stack Software Engineer (Agents & Rapid Prototyping)

Location: Remote (US-friendly hours)
Type: Full-time (founding/early engineer)
Team: Small, passionate, ego-free builders

Why this role

We’re hunting for product–market fit That means we ship POCs fast, test hypotheses daily, and aren’t afraid to throw work away if the data says so. If you code for fun, ask “ why” before “ how,” and love turning messy problems into scrappy automation that actually saves people time—this is your playground.

What you’ll do

• Prototype at lightning speed: Build and ship POCs, internal tools, and production features in days—not weeks.

• Work with LLMs & SLMs: Integrate, evaluate, and fine-tune models; design agentic workflows that reduce manual work.

• Automate the boring stuff: Sniff out bottlenecks, redundancies, and paper cuts; replace them with lightweight automations.

• Balance build vs. buy: Pragmatically choose off-the-shelf services vs. custom code; optimize for speed, cost, and learning.

• Own the stack end-to-end: APIs, data models, frontends, deployments, instrumentation, and on-call for what you build.

• Collaborate without ego: Pair with design, product, and GTM to clarify the “why,” define thin slices, and ship continuously.

• Ruthlessly iterate: Be comfortable restarting weekly if needed; measure outcomes, not lines of code.

Outcomes (how we’ll measure success)

• 2–3 prod-quality launches/month from idea → shipped.

• Clear learning loops each release produces measurable signal (conversion, retention, time-saved).

• Automation ROI: hours saved/team/week; decreased cycle time; reduced ops tickets.

• Agent impact: at least one agent in steady use by internal or pilot users within first 60 days.

You might be a fit if you

• Have 4+ years building full-stack products (or equivalent portfolio/hacker track record).

• Are fluent in a modern web stack (e.g., TypeScript/Node, Python, React/Next; Postgres/Redis; REST/GraphQL).

• Have shipped LLM/SLM features (tool use, RAG, function calling, evals, latency/cost tuning).

• Can design simple, resilient backends; write clean, testable code; and deploy via CI/CD to a major cloud.

• Love debugging with data logs, traces, metrics; you instrument before you guess.

• Communicate crisply, ask sharp questions, and default to action.

• Thrive in ambiguity; you don’t need a JIRA novella to start building.

Nice to have

• Vector DBs, embeddings, prompt/version management, offline evals.

• Agent frameworks (e.g., OpenAI tools, LangGraph, custom planners/executors).

• Queueing/event systems (e.g., Kafka, SQS), WebSockets, streaming.

• Security and privacy basics (authn/z, secrets, PII handling).

• Experience with low-cost scrappy stacks (Cloudflare, Fly.io, Supabase, Firebase).

Our stack (evolving)

• FE: React/Next.js, Tailwind, tRPC/GraphQL

• BE: Node/TypeScript and/or Python FastAPI

• Data: Postgres, Redis/Upstash, optional vector store

• Infra: Vercel/Fly.io/Cloudflare + Docker; CI/CD (GitHub Actions)

• AI: OpenAI, Anthropic, local SLMs when useful; evaluation harness + prompt/version control

How we work

• Learn > Perfect: Ship thin slices, get signal, iterate.

• Write it down: Short RFCs → prototype → user feedback within days.

• Builder culture: No rockstars, no heroes—just ownership and curiosity.

• Time-zone friendly: Async first; quick huddles when needed.

Interview flow

1. Intro chat (30 min): goals, mindset, what great looks like.

2. Builder exercise (home or live, 2–4 hrs): ship a tiny agent or automation; explain trade-offs.

3. Technical deep dive (60 min): architecture, data modeling, reliability, AI evals/cost.

4. Team fit (30 min): collaboration, feedback loops, “why before what.”

Compensation

Competitive salary + meaningful equity. We optimize for ownership and impact