What a position maintaining tutor actually does
A position maintaining tutor is not a bootcamp instructor and not a generic subject-matter tutor. The role exists to keep a working professional employed, productive, and progressing in a job they already hold. The client is not a student trying to land a first role, and not a hobbyist upskilling on weekends. The client is someone who was hired into a position, discovered gaps once the work started, and now needs targeted help to close those gaps before performance reviews, probation checkpoints, or delivery deadlines close in.
The tutoring surface is defined by the job description, the current sprint, and the real artifacts the learner produces at work. If a data engineer is struggling to write production-grade dbt models, the tutoring sessions revolve around the actual repositories, the actual staging tables, the actual PR review comments the learner received last week. If a junior SRE cannot debug a Kubernetes CrashLoopBackOff on the company cluster, the tutor works with sanitized logs and manifests from that cluster. Nothing is hypothetical. Nothing is a toy example.
This makes the work fundamentally different from teaching a syllabus. A position maintaining tutor triages: what is on fire this week, what is due next month, what pattern of mistakes will show up in the next code review. The tutor sequences learning against employment risk, not against a curriculum. The best position maintaining tutors think like a senior colleague who happens to have two spare hours a week to sit with the learner, review their work, and coach them through the parts they are getting wrong.
There are three concrete outputs from a healthy engagement. First, the learner survives probation and hits the 90-day, 180-day, and one-year performance milestones. Second, the learner internalizes the working patterns of a senior practitioner in their domain, so the tutor eventually becomes unnecessary. Third, the learner builds a private evidence file (PRs merged, incidents resolved, dashboards shipped) that can be surfaced in promotion conversations. If any of those three outputs is missing after three months of tutoring, the engagement is failing and needs to be reset.
The framing sits inside the broader position maintaining mentor pillar, which covers the full family of roles (mentor, tutor, coach, advisor) that support employed learners. Tutoring is the most tactical, most hands-on member of that family. It is what you hire when the abstraction of mentorship is not concrete enough and you need someone to sit next to you (over Zoom, with your IDE shared) and unblock the thing that is due Friday.
Why this role exploded in demand between 2023 and 2026
Three structural shifts made position maintaining tutoring one of the highest-growth categories in professional education. The first is the collapse of the traditional onboarding runway. In 2015, a graduate hired into a mid-sized tech company could expect 6 to 12 months of ramp-up: pair programming, low-stakes tickets, structured mentorship from a tech lead. In 2026, that runway is 6 to 12 weeks in most companies, and in some startups it is 6 to 12 days. The gap between what the learner knows on day one and what the job demands by month three is often larger than the employer's onboarding program can close.
The second shift is skill volatility. The stack a data engineer used in 2022 (Airflow, Redshift, Looker) is not the stack in 2026 (dbt, Snowflake or Databricks, Cube or Malloy semantic layers, and increasingly LLM-mediated pipelines). ML engineers who learned scikit-learn and TensorFlow in 2020 are now expected to be fluent in PyTorch, HuggingFace, vector databases, and RAG evaluation. The half-life of a specific technical skill has dropped below the tenure of an average job. People are getting hired into roles that require skills they will need to learn on the job, in real time, without formal training.
The third shift is the rise of AI-augmented workflows that reshape what "competence" means. A junior developer in 2026 is not evaluated on whether they can write a function from scratch. They are evaluated on whether they can prompt a coding agent effectively, review the generated code with judgment, catch subtle bugs the model misses, and integrate the output into a larger system. That is a new skill set, and most universities and bootcamps have not yet caught up. Position maintaining tutors fill that gap because they teach the actual observed workflow of high-performing colleagues, not the workflow described in a 2021 textbook.
Employers, meanwhile, are less tolerant of slow ramp-up. Layoffs across 2023, 2024, and 2025 taught managers that performance-managing someone out is cheaper than waiting. Learners feel this. They know that a bad quarterly review can end an engagement, and they are willing to invest in tutoring to prevent that outcome. The classic argument that "the company should train me" is still true morally, but it is no longer a survival strategy. Serious professionals now treat position maintenance as their own responsibility, and they hire accordingly.
This is also why the position maintainer tutor role has become one of the most consistently requested engagement types on platforms that serve working professionals. Demand is not seasonal. It is not tied to a cohort calendar. It is tied to hiring cycles and quarterly reviews, which run continuously.
How a position maintaining tutor differs from a normal tutor
A normal tutor teaches a subject. A position maintaining tutor teaches a person how to survive and thrive in a specific job. The subject boundary is porous. In a single 60-minute session, a position maintaining tutor might jump from SQL window functions to reviewing a Slack message the learner is about to send to their manager, to debugging a Terraform module, to talking through how to frame a missed deadline in tomorrow's standup. A normal tutor would find that scope creep unbearable. A position maintaining tutor considers it the job.
There are five concrete differences worth naming.
First, the curriculum is emergent. The tutor does not arrive with a syllabus. Instead, the tutor arrives with a diagnostic framework and builds the plan from what the learner shows them. Every session starts with "what happened at work this week" and ends with a small set of committed actions for the next week.
Second, the artifacts are real. Sessions are conducted against the learner's actual repository, ticket queue, review comments, and calendar. No mock projects. No toy datasets. The tutor may have to help the learner sanitize data for confidentiality, but the substrate is always production reality.
Third, the success metric is retention and progression, not test scores. A normal tutor is done when the student passes the exam. A position maintaining tutor is done when the learner has cleared their probation, hit their promotion, or reached a level of independent competence where continued tutoring produces diminishing returns.
Fourth, the pacing is tied to the workplace calendar, not the tutor's calendar. If the learner has a critical launch on Thursday, the Tuesday session becomes a launch rehearsal, and the planned lesson on database indexing gets pushed. Position maintaining tutors are constantly rescheduling their own content against the learner's operational reality.
Fifth, the relationship is confidential and often political. A normal tutor is neutral about the student's other teachers. A position maintaining tutor knows the learner's manager's name, the tech lead's preferences, the team's cultural quirks, and sometimes the political dynamics on the team. That intelligence is used to help the learner navigate, not to editorialize. The tutor is a confidant, not a gossip channel.
For readers who want a side-by-side comparison, we have written a deeper piece on how a normal tutor differs from a position-maintaining tutor that walks through concrete session transcripts and shows the pivot points where the two roles diverge. The short version: a normal tutor optimizes for knowledge transfer, and a position maintaining tutor optimizes for job outcomes.
The diagnostic phase: the first two sessions
The first two sessions of a position maintaining engagement are diagnostic. The tutor's job is not to teach yet. It is to build a map of where the learner stands relative to the demands of the role, and to identify the two or three highest-leverage gaps that will most likely determine whether the learner keeps the job.
Session one is an interview. The tutor asks about the role's official description, the actual daily work (which is often different), the last three tickets the learner worked on, the last piece of feedback from the manager, the last code review with substantive comments, and the learner's own self-assessment of what they are struggling with. The tutor is also listening for signals: how does the learner talk about their team, how much do they know about the systems they touch, do they have a stated understanding of what "good" looks like at their level.
Session two is a work review. The learner shares their screen and walks through recent artifacts: a pull request, a design document, a dashboard, a runbook. The tutor asks questions the learner cannot fully answer, and notes those gaps. This is the moment where the tutor discovers, for example, that the learner has been merging code they don't fully understand, or writing dbt models without a mental model of how the warehouse compiles them, or approving PRs they never actually read.
Out of these two sessions, the tutor produces a written diagnosis. It typically identifies:
- Two or three foundational technical gaps that must be closed within 6 to 8 weeks
- One or two workflow habits that need to change (for example, always running tests locally before pushing, or always reading the full PR before commenting)
- One political or communication risk (for example, the learner has been silent in standups, or has been over-committing on estimates)
- A recommended session cadence, usually 60 to 90 minutes weekly with async support in between
The diagnosis is shared with the learner in writing. This is not optional. Verbal diagnoses drift. A written diagnosis becomes the reference point for whether the engagement is working, and it gives the learner something concrete to push back on if they disagree with the tutor's read of the situation.
A well-run diagnostic phase also sets expectations honestly. If the tutor concludes that the role is a genuine mismatch (the learner was hired for a job they cannot reasonably grow into within the available time), that has to be said. A tutor who takes retainer money knowing the learner is going to be terminated regardless is not doing position maintenance work. They are running out the clock. Good tutors are willing to say "you should be looking for a different role in parallel to this work," and they can point learners toward resources like the position maintaining versus job placement framework to help them think about that decision.
Session structure that produces measurable outcomes
The structure of a position maintaining tutoring session matters more than most tutors admit. Unstructured sessions drift into whatever the learner brought up first, and by the end of an hour the learner has vented but has not moved forward. Highly structured sessions feel efficient but often miss the emergent signals that tell the tutor what is really going on.
The pattern that works consistently is a five-part session, roughly 60 to 90 minutes:
- Status check (5 to 10 minutes). What happened at work this week. Any wins, any incidents, any feedback. The tutor is listening for signals about workload, morale, and any political shifts.
- Commitments review (5 minutes). What did the learner commit to last session, and did it get done. If not, why not, and what does that tell us about capacity or priorities.
- Deep work block (30 to 45 minutes). The core teaching or coaching. This is where the tutor walks through a concept, reviews a PR line by line, pair-programs on a stuck ticket, or simulates a difficult conversation. This block is planned in advance, tied to the diagnosis, and adjusted based on what the status check surfaced.
- Application block (10 to 15 minutes). The learner articulates how they will apply what was just covered, in this week's actual work. Specific tickets, specific files, specific meetings.
- Next-session setup (5 minutes). Explicit commitments (written down by the learner, not the tutor) for what will be done before the next session, and any artifacts to review in advance.
Between sessions, the tutor typically offers async support: a Slack channel or email thread where the learner can drop a code snippet, a screenshot of a review comment, or a question. Async is not "solve my problem." It is a bridge that keeps momentum between the weekly sessions. Good tutors cap async at 15 to 20 minutes of their time per week, and they set that expectation clearly at the start.
We have documented the operational details of this pattern in a dedicated piece on session structure for position-maintaining work, including sample agendas for the first month of an engagement and how to handle the common failure mode where the learner brings a crisis to every session and never gets to the underlying skill work.
One subtle point: the deep work block should not always be teaching. About one session in four should be pure practice, where the tutor observes the learner doing something (writing a query, reviewing a PR, drafting a message) and gives feedback in real time. Practice under observation is one of the most efficient learning modalities available, and most tutors underuse it because it feels like they are not doing enough work. They are doing exactly the right work.
The domains where position maintaining tutoring works best
Not every job benefits equally from this style of engagement. Position maintaining tutoring produces the strongest outcomes in roles where the daily work involves producing artifacts that can be reviewed asynchronously, where there is a clear technical skill layer that can be improved with practice, and where the feedback loop from work quality to employment outcome is relatively fast.
Software engineering is the archetypal fit. PRs, design docs, incident post-mortems, and code review comments are all reviewable artifacts. The skill layer is deep and improvable. Managers can see quality changes within weeks. Whether the learner is a backend engineer working in Go, a frontend engineer in React, or a full-stack developer navigating a Kubernetes-based deployment pipeline, the tutoring surface is rich and concrete.
Data engineering and analytics engineering are equally strong fits. A tutor can review dbt models, look at SQL, walk through Airflow or Dagster DAGs, and coach on data modeling decisions. Tools like Snowflake, Databricks, BigQuery, and the modern semantic layer stack are technical enough to reward structured tutoring but stable enough that the tutor's knowledge does not become obsolete every quarter.
SRE and platform engineering roles benefit strongly because the failure modes are dramatic and observable. A learner who cannot debug a failing ArgoCD sync, or who does not understand how Trivy scan results map to production risk, or who cannot read a distributed trace, will hit visible incidents that create clear teaching moments. The tutor's job is to convert incidents into skill, which requires being available when they happen and having enough operational depth to interpret them.
ML engineering has become one of the largest categories in the last two years. Learners who were hired into ML roles based on coursework often struggle with the operational reality: model deployment, feature stores, drift monitoring, evaluation harnesses, and increasingly the intricacies of RAG systems and agentic workflows. A position maintaining tutor in this space needs to be current with the tooling and honest about the gap between academic ML and production ML.
Security engineering, cloud engineering, and DevOps round out the strongest domains. In all of these, the artifacts are reviewable, the skill layer is deep, and the employment feedback loop is fast enough that the tutoring produces visible results within a quarter.
Roles where position maintaining tutoring is harder include pure people-management roles, sales roles, and heavily relational functions where the artifact is a conversation rather than a document. Tutoring can still help, but the format needs to shift substantially toward role-play, recorded call review, and structured feedback frameworks. A tutor who tries to run a software-engineering session structure with a first-time engineering manager will produce frustration on both sides.
The economics: what learners pay, what tutors earn
The economics of position maintaining tutoring are different from bootcamp instruction or one-off subject tutoring, and understanding them matters both for learners deciding whether to invest and for practitioners considering whether to take this work on.
From the learner's side, engagements typically run 3 to 9 months, with weekly 60 to 90 minute sessions plus async support. Hourly rates for experienced position maintaining tutors in tech domains range from roughly 80 to 250 USD per hour in 2026, with the upper end reserved for tutors who have direct senior experience in the exact stack and company scale the learner is working in. A typical engagement therefore costs between 3,000 and 15,000 USD over its full duration. That sounds like a lot until it is compared to the alternative: a lost job, three to six months of unemployed job search, and the compounding career damage of a termination on the resume.
Learners often ask whether their employer will pay. Sometimes yes, especially in Europe where training budgets are legally protected in some countries, and increasingly in the US where more companies offer learning stipends of 1,000 to 5,000 USD per year. Framing the engagement as "professional coaching" or "technical mentorship" rather than "tutoring" often unlocks these budgets, because the word "tutoring" carries an unfair connotation of remediation. It is not remediation. It is the same category of investment as an executive coach for a senior leader.
From the tutor's side, this is one of the most sustainable freelance categories in professional education. Client relationships are long, retention is high (learners who see progress renew), and the work is intellectually varied because every learner's situation is different. A tutor with 8 to 12 active clients at 90 minutes per week each is running a full-time practice with revenue in the 200,000 to 500,000 USD range annually, depending on rates and location.
The barrier to entry is real, though. This is not work for someone who learned a stack six months ago. Position maintaining tutors need to have done the job themselves at a senior level, ideally within the last two or three years. Learners can tell within one session whether the tutor has been in the trenches, and word travels fast in professional networks. Building a reputation takes 12 to 24 months of consistent work, often starting with clients from the tutor's own network.
For practitioners interested in this path, the entry point is straightforward: become an instructor on Refonte Learning through the application flow, which screens for domain depth and matches accepted tutors with learners whose needs align with their background.
Failure modes and how good tutors avoid them
Even well-designed engagements can fail, and the failure modes are consistent enough to be worth naming. A tutor who has not thought about these in advance will run into them.
The endless status check. The learner arrives every session with a crisis. The tutor spends 45 minutes triaging the crisis and never gets to the skill work. Over six weeks, the learner feels supported but has not learned anything transferable. Fix: cap the status check at 10 minutes, and explicitly name the pattern when it happens. "We have spent the last three sessions on immediate tickets. This week, we are going to spend 45 minutes on your SQL fundamentals even if there is a fire, because if we do not, you will hit this same class of problem again in a month."
The confidence trap. The learner is polite, agrees with everything the tutor says, and appears to be progressing. In reality, they are not applying anything between sessions, and their work is not improving. Fix: insist on written commitments and review them at the start of every session. If commitments are consistently not honored, name it and ask what is actually going on. Usually there is a workload or motivation issue underneath.
Scope creep into therapy. Position maintaining work is intimate. The learner's job security is at stake, and emotions come up. A good tutor holds space for that without becoming a therapist. Fix: acknowledge the emotional weight, but redirect to actionable ground within a few minutes. If the learner is genuinely in crisis (burnout, mental health, family), the tutor should say clearly that they are not the right resource and recommend appropriate support.
Tutor obsolescence. The tutor's own skills become dated. This happens fastest in ML and AI tooling, where the state of the art shifts every few months. Fix: budget 4 to 6 hours per week for the tutor's own continuing education. This is not optional. A tutor teaching 2023 patterns to a 2026 learner is doing damage.
Over-dependence. The learner becomes reliant on the tutor for every decision. This is a failure of the tutor, not the learner. Fix: from session one, communicate that the goal is the tutor's own obsolescence. Push the learner to make judgment calls independently between sessions, and review the calls rather than pre-approving them.
Misaligned expectations with the employer. In rare cases, an employer discovers the tutoring engagement and reacts negatively, either because they feel the learner should be self-sufficient or because they suspect confidential information is being shared. Fix: coach learners on how to describe the arrangement if it comes up. Framing it as "professional coaching" is honest and defensible. The tutor should also be scrupulous about not asking for or storing anything that violates the learner's employment agreements.
Confusing the pillar concepts. Some learners hire a tutor when what they need is a mentor (broader career guidance) or a coach (behavioral change work). The roles overlap but are not identical. Good tutors are honest about when a learner would be better served by a different modality, and they know their own boundaries.
What separates great position maintaining tutors from adequate ones
The difference between a tutor who keeps clients for eight months and one who loses them after two is not primarily technical skill. Both tend to have the domain depth. The difference is a set of practices that compound.
Great tutors write things down. After every session, they spend 10 minutes updating a running note on the learner: what was covered, what commitments were made, what patterns are emerging, what to watch for next time. Six sessions in, the tutor has a rich, structured picture of the learner's trajectory. Adequate tutors rely on memory and lose the thread by month two.
Great tutors ask better questions. When a learner shares a PR that has problems, the adequate tutor points out the problems. The great tutor asks the learner to review their own PR out loud, and lets the learner discover half the problems themselves. That builds transferable judgment. Direct correction only builds specific-instance knowledge.
Great tutors know the difference between "the learner does not know this fact" and "the learner does not know how to think about this class of problem." The first is fixed with a five-minute explanation. The second requires practice under observation and is where most of the tutoring hours actually go. Adequate tutors treat every gap as a fact gap and end up explaining a lot without changing much.
Great tutors are honest about progress. If the learner is not on track after eight weeks, they say so, and they either recalibrate the plan or recommend the learner start a parallel job search. Adequate tutors soft-pedal bad news, hoping the situation will turn around, and by the time it is undeniable, the learner has lost the runway to do anything about it.
Great tutors invest in their own network. They know other tutors in adjacent domains and refer out when a learner's needs are outside their zone. They know recruiters, hiring managers, and senior practitioners they can loop in for specific conversations. Adequate tutors treat the engagement as a solo relationship and never leverage the network that could accelerate the learner's outcomes.
Great tutors also understand the market they are operating in. They read the same job posts their learners are reading. They know what a strong performance review looks like at Anthropic versus at a mid-sized fintech versus at a regulated bank. They calibrate their advice to the specific culture the learner is embedded in, rather than giving generic "best practice" guidance that may or may not translate.
Finally, great tutors treat the work as a craft. They review their own sessions, notice which teaching moves landed and which did not, and refine. Refonte Learning has found that the tutors who consistently produce the strongest learner outcomes are the ones who treat their own practice as something to improve, not as a fixed skill they already have.
How to hire a position maintaining tutor (learner's checklist)
If you are the learner and you are considering hiring a position maintaining tutor, the following checklist will save you months of wasted effort and thousands of dollars.
Start with the diagnosis, not the tutor. Before you look for anyone, write down (in one page) what you are struggling with at work, what feedback you have received, and what specific outcome you need in the next 90 days. This document is the brief. Any tutor who cannot respond concretely to it is not the right hire.
Insist on domain-specific experience. A tutor who has been a senior engineer at a company similar to yours in the last three years is worth 3x one who has taught the subject academically. This is not about credentials. It is about pattern recognition. A senior practitioner has seen the specific class of mistake you are making before and knows exactly how to unwind it.
Ask for a paid diagnostic (one or two sessions) before committing to a longer engagement. A tutor who refuses a paid trial is either overconfident or has not thought about the risk on your side. A tutor who insists on a long minimum commitment is prioritizing their revenue stability over your outcomes.
Ask how they measure success. If the answer is vague ("you'll feel more confident"), keep looking. Good tutors will give you concrete outcomes: cleared probation, promotion within 12 months, PR review comments cut by 60 percent, incidents you can independently resolve. Those are measurable, and they will hold the tutor accountable.
Check references or public work. A tutor should be able to point to at least one or two learners (with permission) who can speak to the engagement, or to public writing, talks, or open source contributions that demonstrate the depth of their thinking. Someone with no external footprint may still be excellent, but the risk is higher.
Be honest about your situation. If you are on a performance improvement plan, tell the tutor. If your manager is hostile, tell the tutor. Withholding information wastes the tutor's diagnostic energy and delays help. The confidentiality goes both ways, and a good tutor will hold whatever you share.
Expect to be uncomfortable. Effective tutoring will surface things you were avoiding: skills you have been faking, feedback you have been dismissing, patterns you have been repeating. If the sessions feel entirely comfortable after the first month, either the tutor is soft-pedaling or you are not actually engaging. Discomfort is not the goal, but it is a common byproduct of real progress.
The Refonte Learning approach and how to get involved
Refonte Learning has invested heavily in the position maintaining category because we believe it is one of the highest-leverage forms of professional education available in 2026. The learner already has the job. They already have the network. They already have the domain access. What they lack is a senior practitioner willing to sit next to them, look at their actual work, and coach them through the specific gaps that are limiting them.
Our model matches learners with vetted tutors who have direct senior experience in the learner's domain and stack. We enforce the diagnostic phase, the written diagnosis, and the structured session format because we have seen consistently that these operational disciplines separate engagements that produce outcomes from engagements that produce billing. We track retention (do learners renew after 90 days), progression (do they hit their stated professional milestones), and independence (does the engagement wind down naturally as the learner grows).
For practitioners who want to teach this way, the barrier is intentional. We interview for domain depth, current stack fluency, and coaching disposition (not everyone with technical depth is a good tutor). The reward for clearing that bar is a stable, well-compensated practice with clients who value what you do, because the value is measurable in their careers.
If you are a senior practitioner considering this work, we would encourage you to apply to teach on Refonte Learning. The application will ask about your specific technical background, the roles and companies you have worked in, and the way you think about coaching versus teaching. It is not a fast process, but it is designed to protect both you and the learners you would eventually work with.
Refonte Learning operates from the UK office at 1 Poulton Close, Dover, Kent, United Kingdom, CT17 0HL, and is registered in France under SIREN 949 841 605 (INPI record at https://data.inpi.fr/entreprises/949841605). Our tutor community spans time zones and domains, and the platform handles matching, scheduling, payment, and quality oversight so that both tutors and learners can focus on the work itself.
The position maintaining tutor is not a new invention. Senior practitioners have been quietly mentoring junior colleagues since the beginning of professional work. What is new is that the role is now formalized, compensated, and available to learners who do not happen to be lucky enough to have a generous senior sitting three desks over. That formalization is what makes the difference between career survival and career loss for a growing number of professionals, and it is why this category will continue to grow through 2026 and beyond.