ENNPRO Get in touch
← All product roles

Product Manager, Applied AI

Decide which parts of an operational workflow are ready for automation.

Apply for this role

About the role

A client may ask for an agent when the underlying need is to clear a queue before the end of a shift. You’ll work with operators and engineers to identify that need, agree on a useful first release, and measure whether the change helps.

You’ll work with the person who owns a workflow and the people who carry it out. You’ll agree which decision a model could support, what the current process costs and which mistakes require human review. Engineering needs a pilot narrow enough to test those assumptions.

Working at ENNPRO

ENNPRO is an applied machine learning consultancy. We work alongside client engineers and the people who use their systems. Depending on the engagement, a client might need a service they can operate, a benchmark they can reproduce or enough evidence to decide whether to release a model. Their team should be able to use the result after handover.

You’ll spend time with the people who feel the cost of a slow or incorrect decision. On one engagement that could be the person assigning documents at the start of a shift. On another it could be the supervisor who has to explain why an item was sent back for a second review. Your job is to understand both perspectives before the team commits to building a feature.

Your partners are the client’s workflow owner, ENNPRO’s engineers and the designer or product engineer developing the interface. You’ll own the product brief and the decision it asks the client to make. Engineers own implementation choices. Work together on the boundaries, particularly where an uncertain prediction becomes a recommendation that someone may act on.

Responsibilities

  • Map a real case from intake to resolution. Measure the queue, handling time, sources of rework and escalation path with the operators who know the exceptions.
  • Write the pilot scope with engineering and design. Define eligibility, examples that must stay with a person and the evidence needed before a suggestion reaches an operator.
  • Choose measures that reveal both benefit and harm. Review handling time alongside corrections, repeat contacts and the work passed to another team.
  • Lead the pilot review with the client owner. Turn the evidence into an explicit decision to expand, revise or stop, and document what would justify revisiting it.
  • Prioritize candidate workflows using their volume, cost of a mistake and ease of obtaining trustworthy feedback. Explain why a seemingly attractive feature belongs later in the sequence or outside the current engagement.
  • Define review and override behavior with the people who will operate it. Agree on who can act, what evidence they will see, and how corrections will be investigated when a user and the system disagree.
  • Write a release recommendation that connects product behavior to operating capacity. If a new threshold doubles the review queue, show the owner what that means for a shift before asking them to approve expansion.

Day to day

A normal piece of work might begin with a walk through yesterday’s queue and end with a one-page pilot brief. Use interviews to find the decisions worth measuring. Then ground the story in actual counts and a few cases that colleagues can inspect. A demonstration should help the client answer a question about the workflow, such as whether an operator can verify a recommendation with the evidence available on screen.

You’ll keep the product record lean enough for the team to use. Maintain agreed definitions, unresolved questions, decisions and the reasons for changing scope. When a pilot uncovers a new need, show what it would take to pursue it and what the team would have to postpone.

Projects and first 90 days

A representative assignment

A team wants an assistant to clear a document queue before the end of each shift. Shadow a shift and examine cases that returned the next day. Agree with the owner on a small category where reviewers can check every suggestion during a trial.

The pilot brief should tell an engineer what to instrument and an operator when to intervene. At the review, separate faster first handling from any extra rework later in the week.

A trial can shorten the time spent on a document while increasing the number returned a week later. Choose a follow-up period with the supervisor and agree on who will review those returns. Include that effort in the evaluation so the pilot reflects the total work required to complete a case.

How your first engagement could unfold

First month. Shadow a workflow with the people who perform and supervise it. Establish how work enters the queue, where it waits and which errors matter downstream. Write the current process back to the team in terms they recognize and identify a decision narrow enough for a first pilot.

Second month. Agree on a baseline, eligible cases, acceptance criteria and the conditions for stopping the trial. Review realistic examples with engineering and design. Rehearse a disagreement between the operator and the model so the team can settle who owns the final action before the pilot starts.

Third month. Bring the pilot evidence to the client owner. Show where handling time changed, which exceptions remained difficult and whether work moved to another team. Recommend the next investment and record the evidence that would cause you to reverse that recommendation.

Qualifications

Core experience

  • You’ve owned a product decision involving uncertain predictions or operational data and can explain the evidence available at the time.
  • You can translate an interview into testable acceptance criteria and work with engineers to identify what a pilot can realistically measure.
  • You can describe a time you narrowed or stopped a rollout and communicated the decision to people with different incentives.
  • You have defined a baseline with an operational team and can explain which observations or data gaps changed your understanding of the original request.
  • You can lead a decision with technical and nontechnical stakeholders, make the consequences of a trade clear and put the agreed decision into a brief that engineers can use.

Useful additional experience

  • Experience with decision support, human review workflows or products where recommendations can be accepted, corrected or escalated.
  • Familiarity with controlled pilots, observational data, qualitative research or measuring downstream effects after a release.
  • Experience partnering with a client product owner or accountable operations lead through a change in process and ongoing ownership.

How to apply

Tell us about a pilot you helped shape. What decision was the pilot meant to inform, which measure proved most useful and what happened at the review?

Use one example to explain how you separated the business problem from the requested feature. Tell us whose decision you needed, what evidence you brought and what happened after the pilot review. A short anonymized product brief or a written account of a feature you chose to stop are both useful.

You can use an anonymized example for confidential work. The application form includes space for your account of the project, an optional résumé and an optional link to a work sample. It currently lets you prepare and review a draft in your browser.

Equal opportunity and accessibility

ENNPRO welcomes applicants of all backgrounds. We consider candidates on the skills and experience relevant to the role, including experience gained through different educational and career paths. If you need an accessible version of these materials or an adjustment to the application process, write to contact@ennpro.pro.

Start application