Why Refonte records live sessions in 2026
Refonte Learning records most instructor-led live sessions to serve learners who cannot attend live, to provide accessible on-demand review, and to maintain teaching quality at scale. Recordings let participants revisit complex demos, rewind hands-on walkthroughs, and clarify key decisions made in class. For a platform that teaches AI, data, cloud, devops, and software engineering, the ability to pause and rewatch a code-along or a whiteboarded architecture review is not a luxury. It is a core part of how technical fluency sticks.
Recording also supports quality assurance and consistency across cohorts. It enables peer review and internal calibration for tutors, and it gives learners a fair path to escalate content issues, such as a mislabeled dataset or a misconfigured Kubernetes manifest discussed in class. When we update a curriculum unit, prior recordings serve as a baseline to measure whether the revised flow improves clarity and engagement.
Accessibility is another driver. Transcripts and captions derived from recordings help non-native speakers and learners with hearing differences follow content precisely. Students can search transcripts for terms like vector database, ArgoCD, or dbt, then jump straight to that segment. This reduces note-taking pressure and frees cognitive bandwidth for problem solving.
Recording is also how we align with the expectations we codify in our Refonte live session quality standards. Those standards cover audio intelligibility, screen readability, pace control, and interactive checkpoints. Without recording, we would be guessing about execution quality rather than verifying it.
We acknowledge that recording changes the social dynamics of a classroom. It can make some participants more cautious about speaking or sharing work-in-progress. The goal is to protect learning value without making people feel surveilled. That is why this policy defines clear scope boundaries, strict retention schedules, optional privacy controls, and a transparent process to request edits or redactions when something sensitive is accidentally shared.
Finally, we aim for parity of learning outcomes across time zones and schedules. Many learners balance work, caregiving, or shift duties. Recording mitigates time zone friction and safeguards continuity when unforeseen events occur. If a session is cut short by a local outage, the recording of the make-up session ensures the cohort progresses together.
If you want a program where live learning plus recording-enabled reinforcement are both first-class, the AI Engineering Program is designed around that stack: live expert teaching, recorded rewatch, and project-based practice.
Scope: what we record, what we do not, and how we label it
Scope is the backbone of a recording policy. We record the instructor’s audio and video, the shared screen, Refonte-managed digital whiteboards, and group audio in the main room. Instructors may spotlight a learner’s shared screen only with explicit in-session consent. We capture polls and main-room chat that is relevant to the lesson flow. We do not record private direct messages between participants, even if they are sent during the session. We do not record breakout rooms by default.
We mark every recorded segment with a persistent on-screen indicator. Tutors must verbally announce at the start that the session is being recorded and remind participants where to find the banner. The LMS session tile includes a Recorded label and the planned availability window. If the tutor pauses recording for an off-topic moment, the pause is announced and confirmed.
We follow the conventions detailed in our Refonte live session format explained article. Cohort lectures are recorded by default. Office hours are recorded when the discussion is primarily about shared curriculum material that benefits the broader group. Career coaching circles, peer retrospectives, and wellbeing check-ins are not recorded. For demo-heavy labs, we record the instructor’s walkthrough and the Q&A, but we do not record individual troubleshooting that exposes a learner’s personal files or employer code.
Breakout rooms are opt-in recordable. If a group wants a breakout recorded to preserve a complex whiteboard, the tutor can create a dedicated recorded breakout with only those who consent. The default remains off. When we run design studios, we record the group critique in the main room rather than parallel breakouts.
Screensharing hygiene matters. Tutors and learners must close unrelated applications and remove confidential documents from the desktop before sharing. If proprietary client data appears, the tutor must stop recording, move the conversation to a sanitized example, and resume only when safe. As a safety net, our redaction workflow allows us to blur or cut accidental exposures after the session.
Artifacts linked to the session, like notebooks, slides, and code snippets, are stored in the course workspace separately from the video asset. They are not embedded in the video file itself, which simplifies redaction and versioning without changing the visual fidelity of the recording.
Consent, notifications, and participant choices
Informed participation is the cornerstone of this policy. Consent is collected through the LMS at course enrollment, reiterated on the session page, and signposted in the live room with a visible Recording indicator. Instructors also give a spoken reminder at the start of every recorded session. If the tutor restarts recording after a pause, they state it clearly.
Learners can participate without turning on their camera. Microphone participation is optional as long as you engage in chat or collaborative boards. If you want to ask a question verbally but prefer not to appear on screen, you can keep your camera off or request that the question be taken after the recording is paused. You may also use a non-identifying display name that your tutor recognizes but that does not reveal your full legal name to peers.
Participants can request special handling before the session. For example, if you plan to share a sensitive architecture that cannot appear on video, you can ask the tutor to designate a non-recorded segment or to route the demo into a redacted post-session walkthrough. Tutors can also allow you to submit code or a diagram asynchronously for review outside the recording.
Consent differs slightly across formats. As explained in 1-on-1 mentoring vs cohort sessions, cohort sessions are recorded by default because the content and Q&A benefit the entire group and future participants. Private 1-on-1 sessions are not recorded unless both parties agree that a recording would be helpful for the learner’s ongoing portfolio or skill development. Even when both agree, we limit distribution to the learner and the assigned mentor or tutor.
Minors require guardian consent where applicable. If a learner is under the age threshold in their jurisdiction, enrollment collects guardian authorization for recorded educational services. Guardians may revoke consent at any time for future sessions. Revoking consent does not retroactively delete prior recordings that were made while consent was active, but it stops future recording and can trigger a redaction review if sensitive content is involved.
Our notifications are designed so no one is surprised. If a scheduled non-recorded session must be recorded due to an unusual curriculum need, the LMS session page is updated at least 24 hours in advance, and the tutor reiterates the change at the start. If you join late, the on-screen indicator and a chat reminder keep you informed.
Privacy principles and lawful basis
Refonte Learning approaches recording with a privacy-by-design mindset. We collect only the data needed to deliver and improve instruction. The main room stream is necessary to fulfill the educational contract with enrolled learners, and we rely on legitimate interests to improve teaching quality through limited internal review. Where the law requires consent for recording, we obtain it through clear just-in-time notifications and enrollment acceptance.
We minimize exposure. Breakouts are off by default. Private chats are never included. We avoid capturing detailed personal backgrounds by encouraging virtual or blurred backgrounds, and we promote screen-hygiene checklists. We instruct tutors to prefer sample or open datasets when illustrating ML workflows, and to redact customer identifiers in devops log streams before sharing.
We apply role-based access control so that only people with a direct educational need can see a recorded asset. Learners in the cohort can stream the video in the LMS. Tutors and quality reviewers can access recordings they teach or review. Operations staff who maintain the platform see metadata, not content, unless needed to diagnose a production issue under a documented access request.
The recordings serve defined purposes: replay for enrolled learners, accessibility through transcript generation, internal quality calibration, and evidence for resolving disputes about academic integrity or attendance. We do not use learner video for marketing without a separate, explicit media release. We do not sell personal data.
We design for data subject rights. Learners can request access to their personal data within recordings, ask for corrections to transcript misspellings of their name, or request redaction of incidental personal information that is not pedagogically essential. When a request is granted, we implement the minimal edit that resolves the issue while preserving instructional value for the cohort.
Cross-border handling follows the residency rules we communicate at enrollment. If a recording is processed by a transcription engine, we ensure that the processor follows contractual data protection commitments and does not use the content to train unrelated services. We keep an audit trail of processing activities that touch recorded assets.
This policy is an educational policy document, not legal advice. We update it when laws or platform capabilities evolve, and we document version changes in the LMS policy log.
Storage, encryption, retention, and deletion
We treat recorded sessions as time-bounded learning assets. We store them in a secure content store with server-side encryption at rest and encrypted transport. Access occurs through the LMS with authenticated sessions and short-lived links. Where caching is needed for reliable streaming, cached segments expire quickly.
Retention is defined by session type. Cohort lectures and labs are retained for 180 days after the cohort end date, then deleted. Office hours that mostly answer problem sets are retained for 90 days after cohort end. Non-curricular coaching is not recorded, so retention does not apply. 1-on-1 recordings, when created by mutual agreement, are retained for 60 days unless the learner requests an extension to finish an assignment or portfolio review.
We publish the planned availability window on the session page so you know when a recording will disappear. If a session becomes a canonical resource for a curriculum unit, we may remaster a segment into a durable asset that replaces a prior clip. In that case, the original recording still follows its retention clock, and the remastered clip is stored as a separate, edited learning object with its own retention.
Deletion is an automated batch process that runs daily. When a recording reaches its retention end, it is queued for deletion. The system removes all video renditions, the raw capture, and derived transcripts. Backups that include the asset age out on their own schedule, and we structure backup retention to ensure that within 30 days of primary deletion, no recoverable copy remains in standard backup rotation.
We keep minimal metadata for 12 months beyond deletion to preserve auditability: session ID, tutor name, course code, and the fact that a recording once existed. This metadata does not include video or audio content. If you need proof that you attended a recorded session after the video is gone, the metadata can provide that confirmation.
When a deletion must be paused due to an active academic integrity investigation or a legal hold, we mark the asset with a temporary hold flag. Only authorized compliance staff can set or clear this flag. When the hold is cleared, the deletion process resumes.
As a convenience for learners transitioning between roles or time zones, we allow tutors to extend a specific recording by up to 30 days if more time is needed to complete a capstone handoff. Extensions are granted sparingly and are visible in the LMS so everyone knows the current plan.
If you are exploring programs that blend live instruction with planned replay and clear retention rules, the AI Engineering Program pairs cohort-based practice with predictable access windows for recordings and transcripts.
Access control, distribution rules, and prohibited uses
Access is role-bound and time-bound. Enrolled learners can stream recordings relevant to their cohort directly in the LMS while the retention window is active. Downloading is disabled for learners and enabled for tutors only when an edit or redaction requires local processing. Quality reviewers can view streams tied to the courses they audit. Operations engineers can access a stream temporarily for diagnostics under a time-limited ticket.
We watermark streamed assets with session ID and viewer ID overlayed at low opacity. This discourages unauthorized redistribution and makes takedown investigations more effective if a leak occurs. We log access events to detect anomalies, such as concurrent streams from geographically distant locations on the same account.
Distribution rules are purpose-built for learning. You may take personal notes that quote the instructor. You may capture static screenshots for your own study if they do not include faces or names of peers and are not shared publicly. You may not upload recordings, clips, or transcripts to public platforms, team wikis at your employer, or any third-party repositories. You may not use audio from a recorded session to train a model or feed a transcription engine outside of the Refonte Learning platform.
We enforce a strict non-harassment and non-disparagement expectation in recorded spaces. Editing out a remark that violates community standards does not remove consequences for the underlying behavior. Prohibited conduct is documented along with the edit, and the incident is handled under the code of conduct.
If you believe a recording has been shared improperly, notify support via the LMS. We will verify the claim, issue takedown notices when appropriate, and disable accounts engaged in distribution violations. We can regenerate access keys for affected recordings and rotate watermarks for ongoing cohorts.
For collaborative projects, we encourage learners to refactor insights into sharable artifacts that contain no personal data. For example, summarize a devops troubleshooting pattern as a clean runbook, or re-create a PyTorch example with public data. These artifacts can be shared freely, which preserves knowledge flow without exposing recorded content.
Redaction, blurring, and edits: how we protect participants
Even with strong hygiene, sensitive information can slip into a live conversation. Our redaction workflow makes it practical to fix the recording without losing the educational moment. Learners or tutors can submit a redaction request from the session page. Provide timestamps, describe the content to remove, and state whether a blur, an audio bleep, or a hard cut is sufficient.
We triage requests within two business days. If the edit protects personal data that is not central to the lesson, we target a five business day turnaround for the first updated version. For urgent cases, such as a password exposed on screen, we temporarily unpublish the recording, create a placeholder banner in the LMS, and fast-track the edit.
What can be edited: accidental exposure of names, email addresses, phone numbers, customer identifiers, API keys, IP addresses tied to a private network, or brief remarks about private health or family matters. We can also cut tangents that reveal an employer’s roadmap or confidential margin data. What we do not edit out is pedagogically essential content. If the only way to remove a disclosure is to cut a whole concept that the cohort must learn, we work with the tutor to re-record that concept and splice in a clean segment.
We handle chat carefully. Main-room chat that is central to the Q&A stays, but we can remove lines that include personal identifiers or that were clearly sent in error. Direct messages are not recorded, so they do not need redaction. If a learner’s legal name appears in the transcript but they use a different name in class, we can correct the transcript.
Versioning is transparent. The LMS marks the recording as Edited and provides a short changelog: for example, Blur applied to 00:11:24-00:11:40 to hide email address on terminal prompt. If a segment is cut, we insert a neutral slide that states Segment redacted to protect personal information without naming the person who requested it.
Our redaction tools preserve audio sync and caption integrity. If an edit shifts timestamps, we regenerate captions and update transcript search indices so that links from the course workbook still land at the right concept. We keep the unedited raw file in a separate quarantine during the edit process and destroy it as soon as the edited master is published, unless a legal hold applies.
Tutor responsibilities and learner etiquette in recorded rooms
Tutors set the tone. They begin with a clear reminder that recording is active, name the topic, and outline when recording may be paused. They model screen hygiene by closing unrelated windows, disabling notifications, and using sanitized datasets. They avoid sharing internal dashboards that might flash customer emails or tickets. When demonstrating tools like Trivy, kubectl, or dbt, they scrub prompts that display credentials or URLs tied to private infrastructure.
Preparation reduces risk. Tutors preflight their content, test microphones and cameras, and load sample files that are safe to display. Our public guidance on expectations is covered in the live session language policy, which complements this recording policy by setting respectful communication norms. Respectful language reduces the likelihood of remarks that need removal later.
Learners play a part too. Join a few minutes early to check your audio, adjust your background, and stage the files you might share. If your question involves a proprietary dataset, describe it abstractly or ask the tutor to pause recording before you screen share. Use the hand-raise feature or chat to queue your question instead of interrupting mid-sentence, which keeps the audio clean and easier to caption.
During Q&A, say your first name slowly so the transcript engine captures it correctly, or state that you prefer not to be named on the recording. If you are answering a peer’s question, stick to the technical issue and avoid comments that could be read as personal feedback. Keep humor kind and inclusive. If a conversation veers into personal territory, ask to pause the recording or suggest moving that part offline.
If you are co-teaching, establish signals. A co-tutor can watch the chat, manage polls, and handle recording pauses while the primary instructor shares the screen. This division reduces cognitive load and supports timely interventions if a sensitive screen pops up. After class, the teaching team reviews chat and flags any lines for removal before the recording is released.
Tutors are also responsible for timely redaction requests. If a learner alerts the tutor to an exposure, the tutor files the request the same day and informs the cohort that a revised version will be posted. This transparency preserves trust and keeps the learning schedule predictable.
Technical delivery: capture pipeline, formats, and reliability
Reliable recording depends on a predictable capture pipeline. We capture the main room audio and video at high fidelity and render multiple streaming bitrates so learners on slower networks can still participate in replays. We record at 1080p with screen-priority encoding for code and terminal clarity, and we capture audio at 48 kHz with noise suppression. We validate input levels before class starts to limit clipping.
We separate tracks when possible. The instructor mic, the system audio from demos, and the composite of participant audio are captured on distinct channels. This separation enables targeted edits, such as ducking background chatter or bleeping a single phrase without muting an entire answer. For whiteboards, we prefer digital canvases that export SVG or PNG alongside the recording to keep diagrams legible.
Transcripts and captions are generated after the session. We run a caption pass that adapts to technical vocabulary and supports corrections. Tutors can review and accept glossary suggestions so that terms like PyTorch, Snowflake, or ArgoCD appear correctly. We sync captions with chapter markers that match the course workbook so learners can jump to sections like Data pipeline linting with dbt or Rolling updates on Kubernetes without scrubbing.
Playback happens in the LMS with role-based gating. Streams are protected with short-lived URLs and cannot be embedded outside the LMS. If a learner moves to a different cohort due to schedule conflicts, entitlements update so they can watch the sessions that match their new timeline. Access logs help us diagnose playback issues, such as repeated buffering on a specific segment that might indicate a corrupted rendition.
We test against the live session technical requirements so tutors and learners know what hardware and bandwidth deliver a good experience. If the live platform experiences an outage, we fall back to a backup room and record locally as a contingency. After class, the local backup is uploaded and reconciled with the primary capture so the cohort still gets a clean replay.
We compress mastered files to balance fidelity and size. Code and small text must remain crisp at typical laptop viewing distances. We validate this by spot-checking with 100 percent zoom on terminal windows and IDE breakpoints. If text looks soft, we adjust encoding and rerender. We treat recording quality as part of instructional quality, not a separate concern.
Participant rights: access, corrections, and objections
You have rights over your personal information in recordings. You can request access to a copy of segments where you appear, ask us to correct misspelled names in captions, or ask for removal of incidental personal data that is not central to the lesson. Use the LMS privacy request form associated with the session. We authenticate your request against your account to protect against impersonation.
If you object to a recording on grounds that affect your safety or privacy, you can request that we unpublish the video while we review the claim. We balance this with the cohort’s learning needs. If an edit can make the content safe, we apply it and republish. If safety cannot be achieved without removing the central lesson, we will re-record the segment and provide the cohort with an equivalent replacement.
If you wish to withdraw consent for future recordings, notify us through the LMS. Withdrawal takes effect for sessions after your notice is received. It does not trigger deletion of prior recordings that were made under valid consent, but it stops future capture of your audio and video. We will coordinate with you and your tutor on alternatives, such as engaging by chat or meeting in a non-recorded 1-on-1 slot.
When you need a copy for accessibility reasons, we handle it with care. We may provide a temporary downloadable file encrypted with a password, or a transcript if that satisfies your need. We do not provide raw source files unless necessary to meet an accessibility requirement. Any download is subject to the same distribution rules as streaming.
If you believe a transcript misrepresents your statement in a way that could be harmful or misleading, you can propose a correction along with a timestamp. We will evaluate whether the change aligns with the audio. If it does, we will update the transcript and mark the change in the version note.
Special cases: assessments, proctoring, and employer-sponsored cohorts
Assessments require extra care. Timed coding challenges or design reviews may be recorded for academic integrity or for structured feedback. We announce proctoring conditions up front, including what is recorded and who can see the material. If the assessment includes protected employer IP, we provide sanitized scaffolds so that you can demonstrate skill without exposing proprietary code.
Employer-sponsored cohorts sometimes use custom policies. If your employer funds your seat and requests tighter controls, we publish a cohort-specific addendum that may further limit distribution or disable chat archiving. We do not weaken baseline privacy or extend retention without a defined educational need agreed by participants.
Capstones and portfolio reviews are often valuable to rewatch. If a review includes sensitive client data, we coach learners to substitute generic descriptions. If a live client demo is required, we can schedule a non-recorded slot for that segment and record a second, sanitized walkthrough for the cohort’s benefit.
We do not record disciplinary or pastoral meetings. If a behavioral issue arises in a live session, we document it separately and apply the conduct process without attaching the live recording to the case file unless required to verify a claim. When that happens, we isolate only the relevant timecodes and restrict access to the conduct team.
Changes, versioning, and governance
Policy evolves. We maintain a version history in the LMS and signal meaningful changes at least 14 days before they take effect for new cohorts. Material changes include retention periods, access rights, and consent mechanisms. Minor clarifications that do not reduce your rights can roll out as documentation updates.
We gather feedback from learners and tutors in quarterly reviews. If we see repeated redaction requests around a particular exercise, we fix the root cause in the curriculum rather than relying on edits after the fact. If caption accuracy dips on a stream of ML jargon, we tune the glossary and coaching for tutors.
Governance includes clear ownership. A policy owner on the academics team is responsible for definitions, and a privacy lead oversees data handling. Platform engineering owns capture reliability and access security. Tutors are accountable for in-room practice that aligns with this policy.
Disputes are handled in predictable steps. We acknowledge your message, gather facts, review the relevant segment, and propose a remedy. Remedies include an edit, a re-record, or guided alternatives for participation. We publish a neutral statement in the LMS to keep the cohort informed when a recording is temporarily unavailable.
We back up policy with practice. Audits sample sessions monthly to check that indicators are present, that breakout defaults are respected, and that deletion jobs run as scheduled. Findings drive coaching and system fixes. The goal is not to punish, but to keep our teaching craft aligned with our promises.
How this policy connects to our broader live learning model
Recording is one part of an integrated live learning design. Our quality rubric, tutoring model, and session types all shape how recordings are created, labeled, and consumed. If you are mapping how the pieces fit, start with the Refonte live session quality standards pillar, which explains the interaction checkpoints, pacing, and clarity guidelines that recordings help verify over time.
Format variety matters. A hands-on lab looks different from a debate on software architecture tradeoffs, and that affects what is captured and how it is edited. For a structured view of session types and what learners can expect to see on replay, the Refonte live session format explained guide is the reference.
Participation pathways differ by context. Some learners thrive in public cohort Q&A, while others prefer private feedback. The recording defaults reflect this, which is why our 1-on-1 mentoring vs cohort sessions comparison is useful when choosing the right channel for a particular question.
Language and clarity are foundational for good captions and fair transcripts. If you want to understand how we set expectations for tone and phrasing that also improve caption accuracy and searchability, review the live session language policy. Technical reliability is the other half of a usable recording, and our live session technical requirements checklist shows the inputs that produce legible code and clean audio.
Refonte Learning treats recording as a teaching amplifier rather than a surveillance tool. The goal is to help you master real tools, from Snowflake to PyTorch, with human guidance you can revisit. The controls in this policy are how we keep that amplifier pointed at learning outcomes and nowhere else.
Exercising your choices and getting help
Most actions you can take are in the LMS. You can see whether a session will be recorded, view the availability window, request a redaction, or ask for a transcript correction. If you need a non-recorded slot to discuss a sensitive topic, use the request button on the session page or speak with your tutor at the start of class. Tutors are trained to pause recording for those moments.
If you encounter a recording that lacks the indicator, or if a session labeled non-recorded contains an uploaded video by mistake, report it immediately. We will pull the asset, investigate, and repost a corrected version if appropriate. If an outage prevents the primary capture, look for a notice about a backup upload within 24 hours.
If you want to change how you appear, you can set a neutral display name in the LMS and apply a blurred or virtual background. You can also choose to use chat as your primary channel. If you want your voice off the recording for a given question, ask the tutor to pause. If you want your prior question removed from a published video, file a redaction request with timestamps.
If you are unsure whether a demo is safe to show, assume it is not and ask for a private review. We can help you substitute public data or stubbed credentials and still get your learning objective met. Where employer policy requires a non-disclosure posture, we will respect that boundary and redirect the exercise.
This policy exists to protect your agency while preserving a high-fidelity record of the lesson itself. If you ever feel the tradeoff is not right in a particular session, tell us. We would rather adapt the format than make you choose between safety and participation.
Refonte Learning updates this recording policy as technology, pedagogy, and regulations evolve. If you want to experience our live teaching model in a program that balances real-time mentorship with recording-enabled rewatch for mastery, explore the AI Engineering Program. We look forward to helping you build durable skills with clarity, respect, and control over your learning data.