Staff Full Stack Engineer

Knit•New York, NY

1d•Hybrid

About The Position

Knit is an AI-native consumer research platform that helps brands automate and accelerate primary research. The platform condenses the entire quant + qual research process from weeks into days or hours for over 50 enterprise brands. Knit is redefining how insights teams operate by automating tasks from survey generation to stakeholder-ready reports. The company is expanding its NYC engineering team and is seeking resourceful, deeply technical individuals who thrive in high-ownership environments. As an early team member, the role will significantly influence the platform's core development and contribute to a team that is projected to grow to over 20 engineers within 18 months. The position involves building and orchestrating AI agents, exploring new UX paradigms, and tackling complex problems in applied AI with substantial ownership across the full technology stack.

Requirements

Experience with LLM/agent systems, evals, and prompt tooling; RAG, embeddings, vector dbs.
Deep knowledge of event‑driven systems, streaming, and workflow engines.
Expertise in observability (OpenTelemetry/Prometheus/Datadog), feature flagging, or A/B platforms.
Background in enterprise security and compliance (SOC 2, GDPR)
Bias to action: short specs, small PRs, daily deploys.
In-person collaboration: we’re building the NYC hub together; whiteboards > long threads.
High ownership: you’ll touch product, talk to customers, and influence the roadmap.

Nice To Haves

You’re a 0 → 1 builder who thrives in ambiguity and has shipped products from scratch at a startup or in a hyper-growth phase. You've seen both - early-stage hustle and later-stage scale and know how to build for it.
You’re a systems thinker with deep technical chops: you design clear architectures and write clean, tested Python and TypeScript.
You think like an owner: you spot problems, propose solutions, and build them without waiting for perfect specs. Your impact is measured in business outcomes.
You use AI as leverage. Tools like Cursor, Copilot, or Claude Code are integral to your workflow, helping you ship faster while maintaining a high standard for correctness.
You believe speed and quality aren’t opposites and can articulate the tradeoffs you make.
You communicate crisply. You share your ideas effectively and thrive in tight feedback loops with product, design, and customers.
You’ve got grit: hard problems energize you, context switching doesn’t scare you, and you’re comfortable deleting code.

Responsibilities

Own 0→1 and 1→N features across the stack, from design docs to building high-throughput data pipelines and scalable FastAPI services.
Architect and own the core agent orchestration engine, designing complex, multi‑agent LLM workflows (routing, tool‑use, memory, evaluation) with deterministic fallbacks, guardrails, and auditable actions required by enterprise users.
Build AI‑native, human‑in‑the‑loop UIs. Craft React/Next.js interfaces that let users collaborate safely with agents, with transparency and explainability by design.
Develop the "factory for the factory". Build our experimentation platform: CI/CD for prompts and agents, offline/online evals, drift detection, safety layers, and rollout controls.
Ship fast, safely. Instrument everything for reliability, write concise design docs, add robust tests, and build dashboards and alerts for features launched.
Partner cross‑functionally with Product, AI, Design, CS; talk to customers and SMEs, and turn ambiguous problems into shipped product.
Obsess over performance, optimizing for latency across the system: caching, model selection, distillation, batching, vector search, and infrastructure efficiency (GPU/CPU).
Operate what you build. This includes on‑call ownership, leading incident response, and driving learning through blameless postmortems.
Level-up the team around you. Mentor and collaborate with other engineers to help make the entire team better.