Jobgether is looking for a Director of Engineering.
**About the Company:** Jobgether is an HR-tech platform that, ironically, uses AI for candidate matching. They act as an intermediary, posting a vacancy for their “partner company.” And this is where it gets interesting: a company operating in the AI-native product space is looking for someone for a position that can already be automated today with the help of… well, you get the idea.
**The Pain and the Task:** According to the description, the “partner company” is in a phase of rapid growth. This is that beautiful and terrible period when teams multiply, processes break down, technical debt grows exponentially, and strategy resembles an attempt to assemble a flying airplane. They need an “adult” who will come in, bring order, build processes, mentor managers, and at the same time ensure the product doesn’t fall apart on its way to a bright future. Classic scaling pain. They are looking for a one-man band who will simultaneously be a strategist, psychologist, architect, and overseer.
**The AI Solution:** Instead of hiring an expensive director whose primary time would be spent in meetings and reading reports, one could create a “Digital Director” — a system of several AI agents that would take on 80% of the routine and analytical tasks of this role. This isn’t a direct human-machine replacement, but rather the creation of a system where management functions are delegated to smart algorithms, leaving final decisions and human interaction to people.
***
**Dialogue in the Future Water Cooler Chat:**
— Heard our guys are looking for an Engineering Director? — asks young team lead Max, sipping his coffee.
— I’ve seen plenty of those stories, — chuckles the grey-haired architect, let’s call him Petrovich. — They’re looking for a “visionary leader,” but what they really need is someone to untangle the chaos in Jira and make sure three teams aren’t building the same feature three different ways.
— Well, how else? Someone has to define strategy, review architecture, negotiate with product managers…
— What if I told you that a system could do most of that? Let’s break it down point by point.
***
**Approaches and Tools:**
**1. Strategy and Architecture: “AI Architect”**
* **Director’s Task:** “Define and implement technical strategy,” “influence architecture.”
* **AI Solution:** An agent is created based on a powerful LLM (like Claude 3 Opus or GPT-4), which is “fed” all relevant information: the current codebase (via vector code analysis tools like Sourcegraph), the product roadmap, quarterly business goals, and competitor tech stack analysis.
* **Tools:** LangChain/LlamaIndex for creating a RAG pipeline, vector databases (Pinecone, Weaviate), access to private models via API.
* **Result:** The agent can, upon request (“Suggest an architecture for new feature X, considering our goal to reduce latency by 20%”), generate several architectural solutions with pros and cons, assess technical debt, and even propose a refactoring plan. It doesn’t make the decision but prepares comprehensive analytics for the CTO or lead engineers.
**2. Delivery and Quality Management: “AI Scrum Master”**
* **Director’s Task:** “Ensure high-quality delivery,” “balance speed and scalability.”
* **AI Solution:** Integration of AI agents directly into CI/CD and the task tracker.
* **Automated Code Reviews:** Tools like CodeRabbit or Sweep.ai don’t just find errors; they also suggest fixes in the project’s codebase style. They check pull requests for compliance with standards, test coverage, and potential vulnerabilities.
* **Process Control:** An agent in Jira/Linear analyzes task movement, identifies stalled tickets, predicts sprint deadline overruns, and automatically escalates issues, generating a brief report: “Team A: Task #123 blocked by dependency on Team B. Predicted delay — 2 days.”
* **Building Trust:** Start small. Initially, the agent only leaves recommendation comments in pull requests. Once the team sees that 9 out of 10 of its suggestions are valuable, it can be granted rights for automatic approval of minor changes or blocking merges with critical errors.
**3. Cross-functional Collaboration: “AI Synchronizer”**
* **Director’s Task:** “Collaborate closely with Product, Design…”, “translate strategy into roadmaps.”
* **AI Solution:** An agent that “attends” all meetings (as a transcriber like Otter.ai, but with deeper analysis). After the call, it automatically generates a summary, highlights decisions made, creates action items, and assigns them to responsible parties in the task tracker. It can translate product owner’s “wishes” from business language into technical specifications and user stories.
* **Example:** After an hour-long meeting with product managers, the system outputs: “**Decision:** We are building feature A. **Technical Requirements:** 1. A new endpoint is needed… 2. UI changes… **Tasks Created:** #456, #457 in Jira for the DevOps and Frontend teams.”
**4. Mentorship and Culture Development (the most challenging):**
Here, AI won’t replace human interaction but can augment it. The system analyzes performance metrics (task completion speed, complexity, code quality) and can highlight for team leads both areas for employee growth (“Ivanov often makes errors of type X; here’s a collection of relevant materials”) and successes that deserve public recognition.
**How to validate AI’s work?**
Validation should be highly pragmatic and based on the same metrics used to evaluate a human director.
1. **DORA Metrics:** We measure 4 key metrics before and after implementing the “Digital Director”:
* Lead Time for Changes (time from commit to production).
* Deployment Frequency (frequency of deployments).
* Change Failure Rate (percentage of “unsuccessful” releases).
* Time to Restore Service (time to restore service after an outage).
If these metrics improve, the system is working.
2. **Productivity and Predictability:** We compare the percentage of plan completion for sprints/quarters. The AI system should increase development predictability.
3. **Team Feedback:** Anonymous surveys. “Is it clearer to you what we’re working on and why?”, “Has the number of routine meetings decreased?”, “Do you feel that the code review process has become faster and more objective?”.
Ultimately, when hiring a Director of Engineering, a company isn’t buying a person, but a function — the function of bringing order to chaos. And today, a significant part of this function can be automated, leaving people with what’s most valuable — final decision-making, creativity, and empathy. And that’s a task not for a director, but for a true leader.
Источник: https://www.linkedin.com/jobs/view/4404719440/
