The posting
About TinyFish
TinyFish builds the enterprise infrastructure that lets AI agents use the live web at scale. Our platform unifies four products behind a single API key — Web Search, Web Fetch, Web Browser, and Web Agent — so developers can search browser-rendered pages, turn any URL into clean structured data, run persistent cloud browser sessions that get past anti-bot walls, and execute autonomous multi-step workflows like form-filling and authenticated navigation.
The hard parts are real: sub-250ms browser cold starts, a custom Chromium search engine, anti-bot mechanisms built at the C++ level, and MCP- and CLI-native integrations with tools like Claude Code, Cursor, and Codex. Teams at companies including Google, DoorDash, Amazon, ClassPass, and Grubhub rely on us for use cases like real-time inventory, competitive price monitoring, insurance quoting automation, and autonomous QA.
That means our customers are developers building agentic systems — and the support they need is deeply technical. This role exists to give them exactly that.
The Role
We're hiring a Support Engineer with a strong technical background — someone who can genuinely understand our platform and our customers' hardest issues, even where they don't personally ship the fix. You'll be the person who takes the gnarliest customer cases — a Web Agent that stalls on a specific login flow, a fingerprint getting flagged by a target site, a fetch returning malformed JSON on a dynamic page — and drives them to root cause: reading logs, reproducing the problem, stepping through the browser session, and pinpointing exactly what's going wrong and why.
You sit at the seam between customers and engineering. Your job is to fully diagnose the issue and hand engineering a precise, well-instrumented picture — clear reproduction, isolated cause, relevant logs and traces — so the fix is fast and correct. You'll unblock customers directly with configuration changes, workarounds, and guidance wherever possible, and route true code fixes to engineering with enough depth that they can act immediately. Where the answer is a pattern, you'll turn it into docs, guidance, or diagnostics so the next ten customers never have to file a ticket at all.
What you'll do
- Own hard technical cases end to end — reproduce, isolate, and diagnose issues across the Web Search, Fetch, Browser, and Agent APIs, including auth flows, anti-bot behavior, session consistency, and structured-output correctness; resolve them directly where possible and drive the rest to closure with engineering.
- Debug at depth — read application logs, network traces, and console output; inspect Chrome DevTools Protocol sessions; reproduce with our SDKs (Python / npm); and reason about failures across browser rendering, agent steps, and API layers to pinpoint root cause.
- Be the technical bridge to engineering — hand off precise, well-instrumented bug reports (clear repro, isolated cause, logs and traces) so fixes land fast and correct, and champion the product gaps and reliability issues that matter most to customers.
- Turn support load into leverage — convert recurring issues into documentation, code samples, quickstarts, and diagnostic guidance that deflect tickets and shorten time-to-resolution.
- Unblock customers directly — resolve issues with configuration changes, workarounds, and clear guidance wherever a fix isn't strictly required.
- Help developers integrate — support customers building on our CLI, MCP integrations, and SDKs across Claude Code, Cursor, and similar agentic dev tools.
- Improve the motion — help define SLAs, escalation paths, and the tooling that lets support scale without linear headcount growth.
What we're looking for
- 3+ years in a technical, customer-facing role — support/technical support engineering, developer support, solutions/forward-deployed engineering, QA, or a software/SRE role you moved into support from. What matters is a genuine engineering background and a track record of debugging real systems, not just triaging tickets.
- Strong technical foundation — you can read Python and/or JavaScript/TypeScript, follow a codebase to understand behavior, and reproduce issues in code. Writing production fixes yourself is a plus, not a requirement.
- Deep debugging instinct — you enjoy chasing a problem to its root cause and are relentless about reproducing and fully understanding an issue before it's handed off or resolved.
- Web platform fluency — solid grasp of HTTP, REST APIs, JSON, authentication (cookies, sessions, headers, OAuth), and how modern web pages actually render and behave.
- Comfort with browser internals and automation — familiarity with headless browsers, Chrome DevTools Protocol, Playwright/Puppeteer/Selenium, web scraping, or bot-detection/anti-bot dynamics is a strong plus.
- Customer-facing clarity — you can explain a subtle technical failure to a developer in writing, calmly and precisely, and leave them more capable than before.
- Bias to prevent — when you see the same issue twice, your instinct is to document or systematize it away.
Nice to have
- Experience supporting or building AI agents, LLM tooling, or MCP-based integrations.
- Familiarity with Claude Code, Cursor, Codex, or other agentic coding tools.
- Background with cloud infrastructure, containerized browser fleets, or high-concurrency systems.
- Exposure to C++ or lower-level systems (our anti-bot and browser layers live here) — enough to read and reason about them.
- Prior experience in a fast-moving startup where the support function was still being built.
What success looks like in the first 6 months
- You independently diagnose and drive to closure the majority of complex technical cases without needing engineering to re-investigate from scratch.
- Time-to-resolution on hard cases is trending down because your diagnoses are precise and reproducible.
- A meaningful share of repeat issues has been eliminated through docs, guidance, and diagnostics you created.
- Engineering trusts your bug reports enough to act on them immediately.
