Infrastructure Engineer — The Token Company

Type: Full-time | On-site | San Francisco, CA Compensation: $150,000–$250,000 + 0%–1% equity Hiring count: 1 Visa sponsorship: Yes — H-1B, O-1, OPT Reports to: Founder (solo founder; name not provided on role page)

The Token Company does LLM interpretability and context optimization research, building custom machine learning models that analyze and compress token contexts before they reach the underlying model. The result is roughly 50% inference cost reduction, lower latency, and measurably higher accuracy for the enterprises and scale-ups integrating LLMs into their products. Seven months old with roughly 1,000 customers, the company raised $11.7M from First Round Capital and Y Combinator, with additional backing from the founders of Hugging Face, Slack, and Dropbox, and has been through both YC and HF0.

Founded: 2025 | Team size: 1–10 (currently ~4 people) | Total funding: $11.7M (Seed) Industry: AI Tools / LLM infrastructure Website: https://thetokencompany.com Office: San Francisco, CA (in-person, hacker house)

• Sole infra owner, full stack, day one: You own every layer of a multi-region GPU stack (AWS, GCP, Base10, on-prem) end to end — not one narrow slice — sitting directly in the critical path of ~1,000 customers.
• Strong backing and early traction: $11.7M raised from First Round Capital and Y Combinator, plus the founders of Hugging Face, Slack, and Dropbox; ~1,000 customers within seven months; YC and HF0 alumni.
• Comp and lifestyle support for high output: $150K–$250K + up to 1% equity, with housing and food provided at the SF hacker house, visa sponsorship, laundry/cleaning, infinite DoorDash, and health/dental — built around a 996 pace.

• No intake transcript captured — an intake video is posted on the Contrario role page but was not transcribed. Points below are drawn from the role body and outreach template, not a call.
• Single hire: one Infrastructure Engineer to own the entire infra stack end to end (not a narrow ML-training role).
• Stack spans multi-region GPU deployments across AWS, GCP, Base10, and on-premise enterprise environments; infra is in the critical path of live customer traffic.
• Team is ~4 people, moving fast; role reports to the solo founder.
• Environment is explicitly 996 (9am–9pm, six days a week), in person in SF.

Own the full multi-region GPU infrastructure stack end to end as the sole infra hire — global low-latency serving, multi-cloud and on-prem deployments, reliability, and cost efficiency — for a seed-stage LLM context-compression company. In-person in SF at a 996 pace.

• →Own the full infrastructure stack end to end across multi-region GPU deployments, AWS, GCP, Base10, and on-premise enterprise environments.
• →Build and maintain super low-latency GPU serving infrastructure that sits in the critical path of live customer traffic.
• →Manage multi-cloud deployments including AWS Marketplace integrations and cloud provider relationships.
• →Design and iterate on deployment, scaling, reliability, and cost-efficiency systems as the sole infra owner.
• →Support on-premise deployments for enterprise clients and ensure performance and reliability at each site.
• →Research and adopt new infrastructure solutions continuously as the stack and customer base grow.

Tech stack: AWS, GCP, Base10, Terraform, Docker, CI/CD, GPU/ML inference infrastructure, AWS Marketplace

• Own cloud systems serving compression API end-to-end
• Build and operate global low-latency high-throughput GPU ML inference infrastructure
• Work with AWS, Terraform, Docker and CI/CD
• Have built and operated production infrastructure at a startup or larger company
• Learn new solutions and technologies quickly
• Improve and research infrastructure solutions continuously
• Based in or willing to relocate to San Francisco to work in person at the hacker house
• Willingness to work startup hours in a 996-style environment (9am–9pm, six days a week)

• Quick learner who grasps products and systems fast
• Experience building for performance and reliability at scale
• Research and product focus mindset
• High ownership mentality
• Startup-minded operator who prioritizes learning and growth over work-life balance
• GPU infrastructure experience in production
• First infra hire at a startup
• Background at an infrastructure company

• Infra scope limited to model training pipelines only
• 20+ years of experience with a slow-moving, process-heavy background
• Prioritizes work-life balance as a primary requirement
• No production infra ownership

• Salary — $150,000–$250,000
• Equity — 0%–1%
• Recruiter payout / bounty — $23,300–$38,800 (15.5% of first-year salary), net 30-60-90 day payout <-- INTERNAL ONLY
• Experience — 2–6 years
• On-site policy — In person in SF (hacker house); 996 pace — 9am–9pm, six days a week
• Visa sponsorship — H-1B, O-1, OPT
• Employment type — Full-time
• Location — San Francisco, CA
• Benefits — Significant equity, housing & food (SF hacker house), visa sponsorship, laundry/cleaning, company off-sites, infinite DoorDash, health & dental

Contrario "Required Candidate Q&A" on the submission form (only first 3 shown on the page — "Show More" not expanded; confirm remainder):

1. Phone number
2. Are you allowed to work in the United States?
3. LinkedIn / Personal website

Stage 1 — Pending Approval — Candidates awaiting initial approval. Stage 2 — Otso screen — Initial screen. Stage 3 — Second person technical — Technical interview with a second team member. Stage 4 — Whole team — Full-team interview. Stage 5 — Take home / Work trial — Practical exercise or work trial. Stage 6 — Offer Extended Stage 7 — Candidate Hired — Candidate accepts and starts.

• Experience range is 2–6 years (structured Job Details field; outranks the role body's "2+ years"). Out-of-range experience is a scaled Red Flag, not a rule-out.
• The 996 pace and SF in-person/relocation requirement are treated as Requirements (per the role body), not just role constraints — a candidate who signals work-life-balance priority (also a Red Flag) or an unwillingness/inability to relocate to SF fails a Requirement. Note the location exception: applying to the role is implicit willingness to relocate, so don't penalize applicants on location; for sourced/cross-subbed candidates who didn't apply, relocation is an outreach question.
• Sponsorship is available (H-1B, O-1, OPT), so work authorization is not a hard blocker here — but confirm status in outreach/on the call.