技术联合创始人 (CTO) - 汽车零售人工智能
Technical Co-founder (CTO) - AI for Automotive Retail
职位机会
每家二手车经销商的利润在他们买下车辆的那一刻就已决定——而不是在出售时。这个决策依赖于直觉、电子表格以及谁有空闲时间，而整个价值超过5000亿美元的行业尚未真正实现现代化。我们认为这是一个等待构建的AI原生领域，而我们是第一个。
问题和解决方案已经通过一位每天亲身经历该问题的联合创始人验证过。
创业项目
Atrium是一个持续运行的AI工作团队，而不是一个仪表盘——这些代理能够观察、学习，并通过自身的表现获得自主行动的权利，部署在经销商集团现有的工作流程中。同样的决策逻辑必须出现在所有现有工作场景中——网页、桌面、我们的浏览器、移动端——因此工作团队跟随人员，而不是相反。一个大脑，覆盖所有界面。
这不是一个周末就能完成的CRUD任务。想象一下数十个专业代理被组织成少量协调的工作流，共享同一个记忆，每个代理都根据自己的表现记录获得自主行动的权利——同时在四个界面上运行。如果这听起来像是一个有趣的问题而非可怕的问题，请继续阅读。
如果正确构建，这将是一个真正的技术壁垒，而市场仍处于开放状态——目前还没有主导的AI原生玩家占据这个领域。关于产品、模型以及我们在商业上的位置的具体信息最好通过对话分享；我们还处于早期阶段，并有意在市场上保持低调。
合作模式
CEO和行业联合创始人已经到位；问题已经验证。我们正在寻找一位技术联合创始人来从第一行代码开始负责产品和工程——拥有完全的决策权，创始人的股权，以及在每次重大会议上的真实席位。
这是一个联合创始人角色，而不是CTO的招聘。作为回报，你将获得创始人的股权，对产品、技术、招聘、融资和战略的共同决策权，以及与CEO在每个重要决策上的真实合作关系。
你将负责
作为联合创始人兼CTO，你将接管生产代码并继续构建产品（第一行生产代码），设定技术方向，并全程领导执行。
- 负责核心架构决策，影响其他所有方面——代理如何在经销商的世界中看到并行动，以及一个系统如何在网页应用、桌面应用、我们的浏览器和移动端保持一致性，而不需要四个独立的产品各自偏离
- 设计自主模型：代理如何通过自身的表现获得行动权利
查看英文原文
The Opportunity
Every used-car dealer's profit is made or lost the moment they buy the car — not when they sell it. That decision runs on gut feel, spreadsheets, and whoever has a spare hour, inside a $500B+ industry that hasn't meaningfully modernized how it buys. We think that's an AI-native category waiting to be built, and we're going first.
The problem and solution are validated with a domain co-founder who lives this daily and has already hand
The Venture
Atrium is an always-on AI workforce, not a dashboard — agents that watch, learn, and earn the right to act on their own, deployed inside dealer groups' existing workflows. The same decision logic has to show up everywhere the work already happens — web, desktop, our own browser, mobile — so the workforce follows the person, not the other way around. One brain, every surface.
This isn't a weekend's worth of CRUD. Think dozens of specialist agents organized into a handful of coordinated workflows, sharing one memory, each one earning the right to act on its own track record — across four surfaces at once. If that sounds like a fun problem rather than a scary one, keep reading.
A real technical moat if built right, and a wide-open field — no dominant AI-native player has claimed this category yet. Specifics on the product, the model, and where we stand commercially are best shared in conversation; we're early and staying deliberately quiet in the market.
The Partnership
CEO and domain co-founder are in place; the problem is validated. We're looking for a technical co-founder to own product and engineering from the first commit — full authority, founder equity, a real seat at the table on every major call.
This is a co-founder role, not a CTO hire. In return for that level of commitment, you receive founder-level equity, founder-level authority with co-decision rights on product, technology, hiring, fundraising, and strategy, and a genuine partnership with the CEO on every material decision.
What you'll own
As Co-Founder & CTO, you will take over the production code and continue building the product (the first lines of production code), set the technical direction, and lead execution end-to-end.
- Own the core architecture decision that shapes everything else — how the agents actually see and act in a dealer's world, and how one system stays coherent across a web app, a desktop app, our own browser, and mobile, without four separate products drifting apart
- Design the autonomy model: how an agent earns the right to act alone, and how that trust is never assumed
- Build the integration layer into a data ecosystem that won't always cooperate
- Build partnerships with ecosystem data providers and third-party systems to land unconventional integrations no standard API gets you
- Own unit economics as an architecture problem, not an afterthought — this many agents, across this many surfaces, has to add up
- Own the design and layering of the system for the dev team, and keep the build matched to what the business and sales side actually need
- Be the technical voice with dealers, design partners, and investors
- Hire and lead the founding engineering team
Co-Founder Profile
Must Have:
- Has shipped agentic or LLM systems into production where being wrong actually costs something
- Has shipped desktop and mobile apps connected to a larger web platform — comfortable with one product living across multiple facets and surfaces, not just a single app
- Has shipped LLM-based products, but also worked with SLMs and SLM fine-tuning
- Has dealt with the real nuances of web scraping — anti-bot defences, rate limits, brittle DOMs, the gap between "works in a demo" and "works at scale against a site that doesn't want you there"
- Has built complex, modelled data graph layers for intelligence — not just a data pipeline, but a structure that reasoning can actually run on
- Comfortable owning infra, data, and product alone — then hiring past yourself
- Treats guardrails and reversibility as core product, not overhead
- Good to have: edge computing and power-constrained systems, high-frequency distributed computing infrastructure
Great to Have:
- Experience building against locked-down or adversarial third-party data
- Vertical/operational domain grit (Automotive)
- Pragmatic and creative — a solution finder who gets to "working" fast and iterates from there
- Co-operative, strategic, and hands-on in equal measure
How to Apply
Please submit your resume, LinkedIn profile, and a brief note on why this venture aligns with your goals as a founder. We will move quickly for the right candidate.
FutureSight is committed to diversity, equity, and inclusion. We welcome applicants of all backgrounds and experiences.
Originally posted on Himalayas