Nango, the Y Combinator W23 startup that sells product-integration infrastructure to developers, has posted an open call for hires across engineering, forward-deployed engineering, marketing, and developer relations. The careers page is short on specifics: no headcount, no salary bands, no named roles beyond the four function labels in the title. What it does say is more interesting than a job board. The company describes itself as “a veteran team of dev-tool lovers, serving developers out in the open, with humility,” operating fully remote, open-source, and focused on “DevExp to high scale to API fragmentation.”
That last phrase is the tell. API fragmentation is the unglamorous problem underneath every AI agent demo that works in a sandbox and falls apart in production. Nango is not hiring to train models. It is hiring to make the connections between models and the rest of the software world less painful, and the forward-deployed engineering role is the clearest signal of where the money and the pain actually sit.
The FDE role is the interesting one
Forward-deployed engineers are the people a vendor sends to sit with a customer and make the product work against that customer’s specific mess. Palantir popularized the model. OpenAI, Anthropic, and a growing list of AI infrastructure companies have adopted some version of it because the gap between “the API returns JSON” and “the customer’s workflow runs” is where deals die. Nango putting FDE in the same breath as core engineering says the company expects integration work to be bespoke at the edge even if the platform is standardized in the middle.
That is a real admission about the state of AI tooling. If integrations were solved, you would not need people on site to wire them. You would need docs. The presence of FDEs implies that every customer has a slightly different stack of SaaS tools, auth schemes, rate limits, and webhook quirks, and that a generic connector library gets you 80% of the way and no further. Nango’s pitch is that it owns that 80%. The FDE owns the 20% that keeps the contract signed.
The careers page also lists “engineers from” a set of unnamed companies as social proof. Tessera could not verify which companies those are from the fetched page, since the logos do not render as text. {/* TODO: verify the specific companies named in the “Work alongside engineers from” logo strip on nango.dev/careers */}
Why this matters for the AI economy
The model layer is consolidating. A handful of labs ship frontier models, and the price per token keeps falling. The differentiation is migrating up the stack to the things that make a model useful inside a business: retrieval, tool use, auth, audit logs, and the connectors that let an agent read a CRM or write to a ticketing system. Nango sits in that layer. So do Composio, Merge, and a cluster of other integration startups. The competition is not about who has the smartest model. It is about who has the most reliable pipe.
That reframes what “AI company” means. Nango does not appear to train anything. It is an AI-adjacent infrastructure play, and the hiring page reads like a developer-tools company from 2019 with the word “AI” quietly implied. The roles are DevRel, marketing, engineering, FDE. No research scientists. No GPU cluster operators. The bet is that the durable business in AI is not the model but the connective tissue, and that connective tissue is a people-heavy, customer-heavy business, not a compute-heavy one.
There is a counterargument worth naming. If agents get good enough at writing their own integration code, the value of a pre-built connector library falls. An agent that can read an OpenAPI spec and generate a working client in minutes undercuts the whole category. Nango’s answer, presumably, is that generated code still needs auth, retries, rate-limit handling, and a place to live, and that the platform provides that. But the company is hiring FDEs, not just platform engineers, which suggests it does not fully trust the automation story either. The human in the loop is still the integration.
The DevRel hire is a distribution bet
Developer relations at an open-source company is not a nice-to-have. It is the top of the funnel. Nango’s page leans hard on “open-source” and “serving developers out in the open,” and a DevRel hire is how you convert a GitHub repo into a pipeline of companies that will eventually pay for the hosted version. The marketing hire alongside it points to the same motion: content, community, and the slow build of trust that developer infrastructure requires.
The interesting question is not whether Nango can hire. It is whether the integration layer stays a business once agents learn to write their own glue code.
The “fast-growing revenues & usage” line is unverifiable from the page. Tessera has no numbers, no ARR figure, no customer count. Treat the traction claim as a claim. {/* TODO: source Nango’s revenue or usage figures if the company will share them */}
What to watch
The forward-deployed engineering role is the one to track. If Nango fills it and the job stays a permanent part of the org chart, the integration layer is a services business wearing a platform costume, and the margins will reflect that. If the FDE role shrinks over time as the platform gets better at handling edge cases, the platform thesis wins. Either way, the hiring page is a small, honest window into where an AI-adjacent startup thinks its hard problems are. It thinks they are in the pipes, not the model. For anyone building agents that touch real business software, that is probably right.