Converge's First Platform Engineer Search in NYC Is a Read on Where Autonomous Ad Agents Strain

Job listings rarely tell you much about infrastructure. This one does. Converge has opened a search for its first platform engineer, an in-person seat in New York City, and the description states the premise plainly: the advertising agents the company sells are expanding faster than the systems that feed them. The posting asks for 4+ years of experience and prices the role at $180K to $240K plus 0.3% to 0.5% equity.

Read alongside the rest of the listing, that is not a routine backfill. It is a founding-level platform hire at a company whose product is already running in production for consumer brands, and whose own numbers describe data volumes most early-stage teams never reach. What follows follows the posting's logic in a different order: first what the agents already do, then what the platform is said to carry, then fit, pay, and process — closing with a data table and a set of questions readers are likely to ask.

Baseline: what the agents already run

Converge builds agents that buy and scale advertising for consumer brands such as Hims & Hers. The product automates the growth loop end to end: launching campaigns, raising spend, learning from results, and producing new ads so the cycle closes without a human hand on every lever.

Today, by the company's own account, those agents make over 90% of all budget changes in the ad account, while managing $1M+ in daily budgets. The posting adds two further claims about behavior rather than volume: the agents learn from user feedback, and they proactively search for more work to automate. For an industry conversation that still mostly revolves around copilots and drafting tools, a system entrusted with the great majority of budget decisions is a meaningful shift in where human attention sits.

Company, traction and capital: the reported figures

The organizational detail in the listing is as notable as the technical one. Converge is described as a team of 13, working in-person in New York City, with the incoming hire becoming the first platform engineer on staff. Commercially, the company reports $3M in ARR and says it serves 200+ customers. On the financing side, it has raised $5.7M from Y Combinator, General Catalyst, and the founders of PostHog and Algolia.

Dimension	Figure stated in the posting	Why a general reader should note it
Headcount and location	team of 13, in-person in New York City	The opening is the first platform engineer, i.e. the first dedicated infrastructure seat.
Engineering base today	3 product engineers	Platform work is currently unowned rather than distributed.
Revenue	$3M in ARR	Agent output is already tied to a recurring commercial line, not a pilot.
Customer base	200+ customers	Multi-tenant pressure on the same pipeline, not a single design partner.
Capital raised	$5.7M	Backers named as Y Combinator, General Catalyst, and the founders of PostHog and Algolia.
Automation share	over 90% of budget changes	The decision layer, not just the suggestion layer, is delegated.
Spend under management	$1M+ in daily budgets	Failure modes are financial and immediate.

What the posting says the platform already withstands

The listing makes an unusual argument about maturity: that a group of just 3 product engineers already operates systems most companies do not touch until much later. Four load figures are offered as evidence.

• It handles $6 billion in annual revenue.
• It ingests 20 TB every month.
• It sustains 10,000 jobs / s at peak.
• Queries join and aggregate 100s of GBs in seconds, fully real-time.

Taken together, these describe an architecture where ingestion, scheduling and interactive analytics share the same operational envelope. That combination is exactly where agent products tend to hurt: the models are cheap to iterate on, but the data paths behind them have to stay correct while spend is moving in real time.

Named projects on the roadmap

Rather than a generic responsibilities block, the posting lists concrete problems the hire is expected to take on:

• Build out services that can ingest up to a trillion jobs / month.
• Examine esoteric Clickhouse CPU and memory flamegraphs.
• Figure out how to materialize the attribution dataset to scale joining and aggregating billions of rows in realtime.
• Build a SQL dialect so agents can query data directly in Converge.

The fourth item deserves a second look from anyone following agent architecture: a query language built for machine consumers, not human analysts, moves the interface between model and data from an API call into something the agent can reason with directly.

How the seat differs from a typical SRE or backend job

The posting is explicit that this is not a standard SRE, systems, or backend engineering role. The engineer is expected to drive their own roadmap, go deep on the hardest infrastructure and systems problems, and pair with product engineers to ship product that scales to the world's largest consumer brands.

A second layer is the working method. The company states that the role expects self-automation with AI, and that coding agents are used for everything — hillclimbing benchmarks, profiling queries, monitoring, debugging incidents, and writing code. In other words, the platform engineer is being asked to industrialize their own workflow with the same class of tooling the product sells.

Three exclusions the posting states outright

1. Not for someone who prefers to architect systems over building them: the engineer will implement everything they design, and should expect to be coding, a lot.
2. Not for someone who wants to own big tech style, multi-quarter migrations: the team works with daily and weekly timelines.
3. Not for someone who prefers working remote over in-person: the team works from its NYC office.

The skill profile the posting prioritizes

• Very strong coding skills, on a stack that is mostly Python, with familiarity with more performant languages expected.
• Deep understanding of databases, both OLTP and OLAP, with Postgres and Clickhouse.
• Experience at scale: seeing firsthand where things fall over in data-heavy systems.
• Pragmatic and customer-minded thinking, on the grounds that the best solutions require understanding what customers need and what is not needed.

The pairing of Python with Clickhouse and Postgres, alongside the flamegraph and materialization tasks listed earlier, describes a role weighted toward query and storage performance rather than orchestration plumbing.

Pay, equity and benefits

Stated as: Salary: $180K - $240K + equity: 0.3% - 0.5%. Private medical, dental, and vision insurance is included, along with pension and 401(k) contributions.

How the hiring process runs

1. Intro call 15 min: Converge learns about motivations, determines why the candidate would be a great fit, and answers questions.
2. Founder chat 45 min: the founder goes into past experiences to learn what motivates the candidate and what it is like to work with them.
3. Technical 1h: candidate and team work through a typical infrastructure problem faced at Converge.
4. Superday 1 day: the candidate joins for a day to build something, meet the team, and be evaluated by it; the day is fully paid.

The posting adds a speed note: this process can all be done in 2 days, and if a candidate wants to move quickly, Converge says it does too — citing that its founding engineer was on a plane to meet the company just days after the first call.

Questions this listing raises for a general reader

Is the role open to remote candidates?

No. The posting describes an in-person New York City position, with the team working from its NYC office.

How much of the advertising workflow is already automated?

According to the company, over 90% of all budget changes are made by the agents, across $1M+ in daily budgets.

How big is the team behind it?

A team of 13 in total, of which 3 product engineers currently carry the engineering function; the new hire would be the first platform engineer.

What is the commercial scale behind those claims?

The company reports $3M in ARR and 200+ customers, with $5.7M raised from Y Combinator, General Catalyst, and the founders of PostHog and Algolia.

Which technologies does the role assume?

A stack that is mostly Python with more performant languages expected, plus Postgres and Clickhouse, OLTP and OLAP fundamentals, and — on the roadmap — building a SQL dialect for agent access.

How long can the process take?

Four stages: an intro call of 15 min, a founder chat of 45 min, a technical session of 1h, and a Superday of 1 day that is fully paid. The company says the sequence can be compressed into 2 days.

Editorial takeaway

Set aside the hiring mechanics and the listing reads as a small, well-documented snapshot of where agent products currently pinch. The automation figures are on the decision layer — over 90% of budget changes — while the load figures are on the data layer, and the gap between the two is precisely what a platform engineer is being bought to close. The roadmap items make that concrete: scaling ingestion toward a trillion jobs / month, materializing attribution so billions of rows can be joined in realtime, and giving agents a native way to query the warehouse. Whether Converge's own execution matches the ambition is a separate question; what the posting offers is one of the clearer public descriptions yet of what autonomous advertising infrastructure says it needs next.