Senior Product Engineer
Open role · Early-Stage AI Infrastructure Startup · New York Metro (On-site)
- Compensation
- $300k – $400k
- Equity
- 0.5%–1.0% equity
- Location
- New York Metro (On-site)
- Work mode
- On-site
The process
- Step 1Application readA recruiter replies within seven days
- Step 2If it is a yesA recruiter contacts you about next steps
- Step 3Company interviewsEach company sets its own
A recruiter replies within seven days. Yes or no.
About the company
Our client is an early-stage, venture-backed AI startup of about ten people, founded in 2024, that builds an AI-native workspace where one knowledge worker can run hundreds of thousands of agents in parallel, processing millions of tokens a minute and compressing days of research into seconds. It has raised about $10M and came through a top-tier accelerator; its backers include the co-founders of several well-known developer and consumer companies.
The role
The engineering team is two people today. You would be the third engineer: a senior generalist who leans backend and can take a project from design to production on their own. The problems are genuinely distributed-systems problems rather than CRUD with a nicer interface — race conditions when a hundred thousand agents write to one data model, real-time streaming of agent output to the client, the trade-off between write throughput and memory per agent, and agents that have to heal themselves. The client is explicit that front-end work is no longer part of the requirement: if you lean one way, backend and distributed systems is the way to lean, and it wants to see the technical complexity of what you built rather than a summary of its business impact.
What you’ll do
- Be the third engineer on a team of two, owning projects from design through to production
- Solve race conditions when on the order of a hundred thousand agents write to a single data model
- Stream agent output to the client in real time
- Trade write throughput against memory per agent, and make the call deliberately
- Build agents that recover from their own failures
- Build user-facing product on top of all of it — product-led engineering, not pure platform work
Nice to have
- Big tech first and then a high-growth startup, with tenures getting longer rather than shorter
- Product-led engineering work on user-facing features, rather than pure platform work
- Go, Redis, publish-subscribe systems, RPC, PostgreSQL, S3 or Temporal in production
- Competitive programming rankings, original code, or other verifiable deep-technical signals
- Experience at a top AI lab, a data-infrastructure company, a top fintech, or a trading firm where you stayed a hands-on builder
Tech stack
Go, with Redis, publish-subscribe messaging and RPC between services, PostgreSQL and S3 for storage, and Temporal for orchestration. Go itself is not required — production programming is, and C and similar lower-level languages are valued.
Compensation & logistics
- $300k–$400k base, plus 0.5%–1.0% equity. The band was raised in late October 2026 for what the client calls top-level, load-bearing engineers
- On-site and full time in the New York City metro area, with no relocation support
- The final-round interview day is a work trial at the company’s San Francisco office
- Open to candidates who need no sponsorship, to visa transfers, and to new visa sponsorships
- Two hires on this role
- Interview process: a 30-minute phone screen with the chief technology officer, a 30-minute technical interview on data structures and algorithms, a one-hour systems-design interview on distributed systems and scale relevant to agent orchestration, then a one-day on-site work trial
- The client wants few, strong candidates rather than volume
Similar roles
Archive photographs: “Seventh Avenue looking south from 35th Street, Manhattan” (1935) and “Hardware store, 316–318 Bowery” (1938) by Berenice Abbott, public domain; “Crosswalk of Market at Third, San Francisco” by Dllu, CC BY-SA 4.0; “View east from Twin Peaks to Market Street” by Tim Adams, CC BY 3.0. All via Wikimedia Commons, shown in greyscale.
