# Zep is hiring three people to build the Context Lake > A careers page that reads like a filter: three roles, one founder, and a hard line against anyone who needs a playbook handed to them. - Title: Zep is hiring three people to build the Context Lake · Getzep - Summary: A careers page that reads like a filter: three roles, one founder, and a hard line against anyone who needs a playbook handed to them. Zep is hiring… - Keywords: hiring, ai agents, startups, memory, enterprise, technology, Zep, three, people, build, Context, Lake - Source: Getzep — https://getzep.com/careers - Author: Zep Software, Inc. - Read time: 3 min - Topics: hiring, ai agents, startups, memory, enterprise, technology ## Memory for every agent Zep builds the Context Lake, infrastructure that lets AI agents remember and reason across chat, documents, events, and business data rather than just the conversation in front of them. > Memory across every user, every domain, every agent. ## The limit is context Most memory systems see only the current thread. They miss the user, the business, and all the work that came before, so the agent starts cold every time. > Agents are limited by the context available to them. ## Governed and temporal The Context Lake tracks how facts change over time and scales to millions of context graphs per deployment, all under governance rules the enterprise sets. ## Who already runs on it Samsung, Zscaler, Twin Health, HoneyBook, and NASDAQ 100 and Fortune 500 tech companies serve context graphs in under 200ms inside their own VPCs. > served in under 200ms, inside their own VPCs ## Graphiti in the open Zep's open-source context graph framework has passed 30,000 GitHub stars, which is the distribution engine underneath the enterprise product. ## Small, distributed, opinionated The team pairs on hard problems, reviews each other's designs, and treats learning as part of the workday. Disagreement is expected out loud, early. > When we find pain, we go fix it. ## One marketer, no team The Marketing Manager runs enterprise B2B demand generation solo at $120K to $180K plus 0.15% to 0.35% equity. There is no team to inherit and none to manage. > a team of one as an advantage ## Buyers ask Claude first The role owns answer engine discoverability, not just Google SEO: structured data, llms.txt, docs architecture, and proof that any of it moved a number. > Buyers ask Claude and ChatGPT before they ask Google ## Founders write, you ship Content strategy and most long-form writing come from the founders. The marketer runs the calendar, adapts each draft per channel, and owns publication. ## Build the agents, keep them You build agents that produce content, run campaigns, watch competitors, and report results. You own them after they ship, including the cost of running them. > You own them after they ship. ## Ninety days to a number Day 30: blog and social running on your agent pipeline. Day 60: funnel measured end to end. Day 90: you own the demand gen number and defend a roadmap. ## Research that ships The Member of Technical Staff role pays $180K to $250K with 1.00% to 1.50% equity to build retrieval agents and finetune the models that extract memory. > You build agents that improve retrieval. ## Papers do not count Zep wants classical ML in real products, finetunes that reached production, and agentic systems that are not thin wrappers over a chat completion API. > Not a prototype, not a thin wrapper over a chat-completion API. ## Write up the failures Results that killed an idea get documented too, so the rest of engineering can build on what did not work instead of rediscovering it. ## Evals catch regressions Part of the job is building eval harnesses for retrieval, memory quality, and agent task completion. By day 90 yours runs on every release. ## Deployment is engineering The Head of Forward Deployed Engineering earns $220K to $270K with 1.20% to 1.75% equity, leading Managed, BYOK, and BYOC installs inside customer infrastructure. > Getting Zep into production in customer environments is its own engineering problem ## The first year is hands-on This leader sits in architecture reviews, sends patches to the codebase in Python, Go, or TypeScript, and shows up on site when a deployment needs them. > You will read our codebase and send patches. ## Build it twice, productize it Success is measured by time from signature to production and by how little custom engineering each deployment takes. Repeated work becomes a reference architecture. > What you build twice becomes a reference architecture or a product requirement. ## Three calls, then done Every role follows the same path: a screening call with founder Daniel, two to three hours of back to back team calls with a walkthrough, then a decision call with Daniel. > We respect your time and keep our interview process tight ## What you get back Platinum medical, dental, and vision, 401K matching, unlimited PTO, a phone stipend, and flexible San Francisco office culture with remote options and travel. ## Key takeaway Zep is hiring operators who ship alone with agents, not managers who assemble teams first.