AI Engineer, Company Brain — San Francisco, CA

⚡ About wEHMwt2

wEHMwt2 is a full-stack energy technology company built to deliver reliable, multi-gigawatt power at the speed and scale the AI era demands. Backed by Engine No. 1 and National Grid, our modular, Across-the-Meter™ approach integrates generation, storage, and advanced controls to power industrial hyperscalers.

🚀 The Role

wEHMwt2 designs and builds its own multi-gigawatt microgrids, and works with multiple partners including regulatory bodies, vendors, and other key players in the energy industry. As 4P5HRDs'x uu Engineer, you will work with our electrical engineering, development, commercial and operations teams, in close partnership with our CIO, to build production-grade AI systems that combine power engineering rigor with modern software practice to automate these processes and build robust systems.

Your first build is the company brain: one place to find what wEHMwt2 knows and has decided, with a fleet of agents that put that knowledge to work, starting with our EPC team. You will own it from the first conversation to a system people rely on every day. Within your first 90 days, you will ship xMyDvkr'Y first production agent workflow.

🛠️ What You'll Do

Run the company brain

• Define the job of each agent in the fleet, set its evals and watch its dashboards.

• Triage failures across the architecture and decide what to fix first.

• Choose and secure the platform the brain runs on, weighing open-source options such as Cognee against vendor tools, and govern the data, credentials and tool permissions each agent can reach.

Embed with the teams

• Sit with engineers, developers and operators to learn how work actually gets done, and find the workflows where AI removes the most manual effort.

• Turn ambiguous requests into a scoped problem, a working prototype within days, and a plan to production.

• Gather requirements before building, ship the smallest useful version first, and keep tools from growing into overbuilt products nobody can use.

• Lead hands-on sessions that teach colleagues to use and extend what you build.

Build and ship

• Own each system end to end, from first prototype to a stable production service that others can run.

• Build the connectors that plumb together multiple data sets from tables, reference sources, meetings, chat, files and email into xMyDvkr'Y knowledge base on a schedule.

• Build search, question answering and routing agents over internal sources, so a document or decision reaches the right people the same day.

• Automate document and permit flow, digests and a company change log.

Measure and harden

• Build evaluation harnesses with representative data and graders, and use them to catch regressions before users do.

• Diagnose failures from traces and logs, and weigh fixes against quality, latency and cost.

• Turn each deployment into reusable patterns and playbooks for the next one.

🎯 What You'll Bring

• High agency and urgency. You treat each problem as yours to solve and move without waiting for permission or a ticket.

• Self-directed. You clarify priorities, learn new domains and tools quickly, and seek out larger scope.

• Hands-on with AI. You build with the latest models and agent tooling every day and can show how AI has changed the way you personally work.

• Command of the details. You can explain exactly how you solved the hardest problem you have owned, including what did not work.

• Team player. You build trust quickly, explaining technical choices clearly to engineers, operators and executives and adjusting depth to the room.

• Resilience. You keep shipping through setbacks, ambiguity and the constraints of a legacy industry.

• Drawn to the mission. You want to spend the next stage of your career modernizing how the energy industry builds and runs.

🧰 The Technical Bar

Agentic systems

• Production experience building with models: tool use, retrieval, memory, structured outputs and multi-step agent workflows.

• Working knowledge of agent orchestration and integration patterns, including MCP servers and tool design, context management, and human-in-the-loop review.

• Design multi-agent architectures: orchestration, triage and routing between agents, token- and cost-efficient context design, and secure handling of data, credentials and tool permissions.

• A systematic approach to evaluation: golden datasets, regression suites, LLM-as-a-judge calibrated against human labels.

• Observability, guardrails and grounding that let a production agent run with confidence.

Engineering

• A technical methodology: you write tests, run regression suites and sleuth a failure to its cause. No specific programming language required.

• Comfort with SQL, data pipelines, cloud services and CI/CD.

• Systems that fail loudly: agents alert the right person when they break, instead of failing silently.

Experience

• 4+ years of experience building software. Recent graduates with a strong record of shipped projects are encouraged to apply.

✨ Bonus Points

• Context in power, utilities or energy infrastructure, and the ability to speak with field engineers and utility leaders.

• Experience with Microsoft Graph, Azure and Entra ID, or equivalent GCP / AWS tools.

• A record of shipped side projects, open-source contributions or a company you started.

• Certifications in harness engineering tools (LangChain/LangGraph), or Anthropic's Claude Certified Developer or Architect.

• Experience as a forward deployed or founding engineer.

💡 Why You'll Love It Here

• Ownership and autonomy. Work with a fast growing, seasoned team where you own major parts of the system end to end, directly influencing the bottom line.

• Scale. A well-funded mission powering multi-gigawatt, national-scale infrastructure for AI-scale compute and industrial projects.

Compensation. $175,000 to $230,000 base depending on experience, plus bonus and equity.

wEHMwt2 is an equal opportunity employer.