Metric Tree Labs
Website:
metrictreelabs.com
Job details:
Location: Remote (India), working with a San Francisco-based founding team
Engagement: Full-time, long-term, dedicated
Experience: 8–10+ years in software engineering, including 2+ years building production LLM applications and agents
Openings: 1
Time zone: Flexible hours, with a daily live overlap with the founders
About the Client
Our client is an early-stage AI startup in San Francisco, founded by alumni of Harvard and MIT. The company is in stealth. It is building an agent-based product that takes complex, high-stakes paperwork off the hands of US families: working out what applies to them, researching official sources, handling correspondence and following up over weeks, with every step checked and backed by evidence.
Parts of the product are already built and running in test. The team today is two co-founders and one engineer. As Lead Engineer, you will be the most senior technical person on the team. You will be the founders’ technical partner, own the architecture, and lead the product into production with real users.
The Role: CTO Judgement, Builder’s Hands
We are looking for someone with the judgement of a CTO who still writes code every day. Expect roughly 70–80% of your week to be hands-on: designing, building, reviewing and shipping. The rest goes into technical direction, standards and growing the team.
• Own the architecture end to end, and be accountable for how the system behaves in production.
• Be the founders’ technical partner: shape the roadmap, make build-vs-buy calls, size work honestly, and push back when something is the wrong thing to build.
• Set the engineering bar for code review, testing, evals, security and releases.
• Lead and mentor the engineers on the team, and help interview future hires.
• Lead from the front in production: when something breaks, you’re on it, and you make sure it doesn’t break the same way twice.
This role is not for you if you haven’t shipped production code in the last 12 months, or you’d rather manage engineers than build alongside them.
What You’ll Own
• Agent platform architecture: design the agent harness, with tool-calling loops that have budgets, stop conditions and failure recovery, and a check on every output before it reaches a user or an outside party.
• Model layer: route across model vendors with fallbacks, manage prompts and versions, and weigh the cost, latency and quality behind each choice.
• Durable, long-running workflows: agent runs that wait days or weeks for a reply, survive restarts and deployments, and can be audited afterwards.
• Evaluation and safety: build the eval framework that decides when an agent is correct, when it can act without a person approving, and when it must escalate. Trace and observe every run.
• Data, security and compliance: store sensitive personal records securely, with row-level access control, encryption, secrets management and audit trails. Get the platform ready for the security reviews (e.g. SOC 2) that customers and partners will ask for.
• Infrastructure and reliability: AWS and Azure architecture, infrastructure as code, CI/CD, monitoring and alerting. Build performance-critical services in Go, Rust or Python where TypeScript is not the right tool.
• Full-stack product: you will still ship features end to end, from the agent to the screen a family sees on phone or desktop.
Milestones
• First 30 days: ship code in week one. Review the existing system, then agree an architecture, a risk list and a 90-day technical plan with the founders.
• By 90 days: the core agent harness, eval framework and production infrastructure are in place, and the first real users are live.
• By six months: the platform runs in production with real users, with measurable reliability and eval coverage, and the team ships to your standard.
Core RequirementsSeniority and leadership
• 8–10+ years of professional software engineering. In recent years you have worked as a tech lead, principal or staff engineer, founding engineer or CTO in early-stage startups, and still written production code throughout.
• You have owned the architecture of a production system from early design through scale, and lived with the consequences of your decisions.
• 0-to-1 experience: you have taken a product from prototype to real users, at an early-stage startup or on a new product line.
• You have led or mentored a small team (2–8 engineers) without big-company process around you.
Agents and AI systems (hands-on, in production)
• 2+ years building LLM applications, including at least one agent system you architected, shipped to real users and ran in production.
• Deep understanding of agent architecture:
– tool and function calling
– planning vs. execution, and single vs. multi-agent designs
– context and memory management
– human-in-the-loop approvals and guardrails
– defence against prompt injection
• You know the frameworks (LangGraph, OpenAI Agents SDK, Claude Agent SDK, Vercel AI SDK, MCP) well enough to know when to write the loop yourself.
• Structured output and failure handling: you get typed, schema-validated output from models, and you handle the cases where the model gets it wrong with retries, repair and fallbacks.
• Evaluation at production level: you have built eval sets and graders (including calibrated LLM-as-judge), wired regression gates into CI, and used the results to decide whether to ship.
• Running models in production economically: caching, routing, batching, and choosing the right model for each task.
• Designing around interfaces, so a model vendor, cloud or database can be swapped without a rewrite.
Engineering depth
• Expert TypeScript across the stack: Node.js on the server and React in the browser.
• Production depth in at least one of Go, Rust or Python.
• Designing and running systems on AWS or Azure, with infrastructure as code (Terraform, Pulumi).
• PostgreSQL: data modelling, query performance and row-level security.
• Durable workflow engines and queues in production (e.g. Temporal, Inngest, Step Functions, Celery).
• Security engineering for sensitive personal data: threat modelling, access control, encryption, secrets handling and audit logging.
• Observability and incident response: tracing, metrics, alerting and post-mortems.
• Automated testing as a habit, including tests of behaviour that depends on a model.
• Heavy daily use of AI coding tools (e.g. Claude Code, Cursor, Codex). You set the team’s standard for reviewing what they produce.
Nice to Have
• Hands-on experience with both AWS and Azure.
• A previous founder, co-founder or CTO role at a startup.
• Research agents that search the web and cite sources; browser automation or voice agents; MCP servers you have built.
• Document processing: OCR and pulling data from scans and PDFs.
• Compliance work such as SOC 2, HIPAA or GDPR.
• Products in high-stakes, regulated areas such as legal, financial, government or healthcare paperwork.
• Public repos, technical writing, talks or open-source work on agent systems. This will make your application stand out.
What We Look For Beyond Code
• A founder mindset: comfortable with ambiguity, owns outcomes, and cares about the people using the product. They are going through a hard time, so every screen, message and automated action matters.
• “Evidence before claims”: when you say something works, you show the test, the log or the screenshot.
• A clear technical communicator: you can explain a trade-off to non-technical founders in plain words, and write a design doc engineers can build from.
• Excellent written and spoken English. You are comfortable on video: a daily stand-up with cameras on, plus short recorded video updates.
• A reliable internet connection, a quiet workspace, and your own laptop (32 GB RAM preferred, disk encryption on).
Working Hours
The founders are in San Francisco, and the wider team works across time zones. You should be open to any of these patterns (India time), agreed with the founders:
• 9:30 a.m. – 6:30 p.m.
• 2:30 p.m. – 11:30 p.m.
• 8:00 a.m. – 5:00 p.m.
Engagement Details
• Full-time, long-term engagement through Metric Tree Labs, working only with this client as part of its core team.
• Target start: as early as possible. The joining date is agreed on selection.
• Senior, competitive compensation, reviewed after six months.
Click on Apply to know more.
