Why Continuing Education Matters
AI evaluation is a field with a half-life of knowledge measured in months, not years. A certification earned in 2023 represents competency in the field of 2023. But the field of 2024 includes new benchmarks, new evaluation techniques, new understanding of fairness implications, new regulatory frameworks. By 2025, substantially new content has emerged. Without active, ongoing learning, certification holders become outdated.
This reality shapes our approach to credential maintenance. Unlike many professional certifications (which require continuing education primarily to maintain market value), the eval.qa credential requires ongoing learning because the field fundamentally changes. Your L3 certification today is meaningful. Your L3 certification three years from now without continuing education would be misleading—it would claim competency in current practices without reflecting that you haven't kept pace with the field's evolution.
Continuing education requirements serve three functions. First, they ensure credential validity: certified evaluators remain qualified in their claim. Second, they maintain professional standards across the field: everyone at a given level has roughly equivalent, current knowledge. Third, they create incentives for the kind of professional development that strengthens the field overall: conference attendance, publication, teaching, collaborative standard-setting.
Research on professional certification shows that fields with structured continuing education have better average practitioner performance and lower rates of ethical violations. Practitioners who engage with ongoing learning networks—conferences, study groups, professional discourse—demonstrate better judgment on novel problems and higher awareness of industry developments. Continuing education requirements operationalize this insight.
The alternative to structured continuing education is informal self-directed learning. Some practitioners naturally stay current through reading, experimentation, and professional networks. But institutional reliance on this is unrealistic at scale. Structured requirements ensure that credential validity doesn't depend on individual motivation, creating a baseline of professional engagement across the certified population.
The EHC System Explained
EHCs—Eval Hours Credits—are our currency for continuing education. One EHC equals one hour of structured learning in AI evaluation (not all learning is structured; more on that below). Your credential maintenance requirements are specified in EHCs; all approved activities are valued in EHCs.
EHC Requirements by Credential Level
The EHC requirements vary by credential level and follow a principle: higher-level credentials require more frequent recertification and more EHCs because expectations for keeping current are higher for advanced practitioners.
- L1 Eval Foundations: 20 EHCs per 2-year cycle. This is the entry level; practitioners are consolidating fundamentals but don't need to maintain bleeding-edge knowledge.
- L2 Evaluation Methods: 30 EHCs per 2-year cycle. Intermediate practitioners should engage with methodological developments and emerging best practices.
- L3 Advanced Practice: 40 EHCs per 3-year cycle. Advanced practitioners have more flexibility (longer cycle) but must engage deeply: 13-14 EHCs annually.
- L4 Exam Design & Leadership: 50 EHCs per 3-year cycle. These practitioners must stay deeply current and often contribute to field development.
- L5 Thought Leadership: 60 EHCs per 3-year cycle. L5 holders are expected to be field leaders; the EHC requirement ensures they're actively engaged with field evolution, not just resting on prior credentials.
These requirements are designed to be achievable for practitioners actually doing evaluation work, not burdensome. A full-time evaluator easily accumulates 15-20 EHCs annually through normal professional activities (attending one major conference, taking a course, publishing). The requirement is a floor, not a ceiling, and most certified practitioners exceed it.
EHC Counting Logic
Not all professional activity is equally valuable for keeping current. The EHC system distinguishes between structured learning (concentrated, focused, with defined outcomes) and ambient learning (exposure to new ideas through work).
Only structured learning generates EHCs. Reading a blog post, even an excellent one: 0 EHCs. Attending a 1-day conference workshop with learning objectives and post-event assessment: 6 EHCs. Completing a 5-week online course: 35-40 EHCs. This distinction ensures that EHC requirements push toward genuine learning engagement, not passive exposure.
The amount of time you invest should roughly match EHCs earned. A 2-hour synchronous workshop = 2 EHCs. A 40-hour graduate course = 40 EHCs. This isn't perfectly calibrated (some activities are more efficient than others), but it's a reasonable guideline. If you're earning more EHCs than hours invested, you're likely inflating claims; if you're investing far more hours than EHCs earned, you may be doing valuable work that doesn't fit the structured learning category.
Approved EHC Activities
The eval.qa community determines which activities generate EHCs. Our approved list spans research, practice, community engagement, and professional development. Here are the major categories:
Notable Gaps and Intentional Exclusions
Some activities that sound like they should generate EHCs don't. We're intentional about this because the EHC system should incentivize activities that directly advance field knowledge, not activities that are valuable but don't fit.
Reading publications: Reading someone else's paper, no matter how important, doesn't generate EHCs. Why? Because we can't verify engagement or learning, and we want to incentivize producing knowledge or formally engaging with it (through courses, workshops, discussions). If you're reading papers as part of a systematic literature review for research, that would fall under research output, which does generate EHCs.
Internal training and study groups: If your company runs internal eval training, it doesn't automatically generate EHCs unless it meets criteria: documented learning objectives, external facilitator or formal curriculum, verifiable attendance and assessment. Casual lunch-and-learns don't count. This maintains standards while acknowledging that not all learning happens through official channels.
Individual tool learning: Learning a new tool (Python library, evaluation platform, cloud service) doesn't generate EHCs directly. But if you publish a tutorial, teach a workshop about it, or contribute to open source in the process, that does. The incentive is to share your learning, not just acquire it.
Emerging Approved Activities
The EHC system evolves as the field does. Recent additions reflect field priorities:
- AI Safety Research Collaboration: 10-20 EHCs. Participation in formal AI safety research related to evaluation—either your own research or collaboration with established safety organizations.
- Regulatory Engagement: 8-15 EHCs. Participation in regulatory consultation processes (EU AI Act implementation, NIST AI RMF development, etc.). Helps the field bridge practice and policy.
- Cross-disciplinary Learning: 5-10 EHCs for completion of substantial courses in adjacent domains (fairness, interpretability, responsible AI) that deepen evaluation foundations.
The eval.qa Continuing Education Committee reviews emerging activities annually and adds new categories based on field developments and member input.
The Recertification Cycle
Your credential has a validity period. When that period ends, you must renew or your credential lapses. Renewal is contingent on meeting EHC requirements.
Cycle Timing and Tracking
Your certification date defines your cycle start. If you earn L2 certification in June 2024, your first recertification is June 2026 (2-year cycle). You must accumulate 30 EHCs by June 1, 2026 to maintain the credential. If you don't, your certification becomes inactive as of June 2, 2026.
The EHC portal tracks your progress. Log in to view your certification details, recertification deadline, current EHC count, and any pending submissions. You can see exactly how many more EHCs you need and when you need them. The system sends reminders at 6 months before deadline, 3 months before, and monthly in the final quarter. This is designed to prevent surprises: you know your deadline, you know your current status, you know exactly what's required.
For practitioners with multiple credentials: if you hold both L3 and L4, you have separate cycles and separate EHC requirements. However, some EHCs can count toward both credentials if they're relevant to both. A paper on evaluation methodology might count toward both L3 and L4 recertification. Document which credentials each EHC supports.
Grace Periods and Extensions
Life happens. You have a 3-month grace period after your recertification deadline. If you miss your deadline by 3 months, you can still renew without additional requirements. Your credential remains active during this period, but renewal is urgent.
Beyond 3 months, your credential becomes inactive. You can no longer claim the certification (you can't list it on your resume, you can't use it for professional purposes) until you renew. However, reinstatement is possible even years later (see lapsed credentials section).
Extensions (beyond the grace period) are available only for documented hardship: medical leave, caregiving obligations, displaced worker status. Request extensions before your deadline through the renewals portal. These are not routine and not guaranteed, but they exist for genuine circumstances beyond your control.
Documenting Your EHCs
EHC claims aren't self-reported honors system. You must provide evidence. This isn't bureaucratic gatekeeping; it's how we maintain credential value. If anyone can claim 50 EHCs with zero documentation, the EHC system is worthless.
Required Evidence by Activity Type
- Conferences: Conference registration receipt, agenda showing your attendance, photos from sessions you attended, or signed certificate of attendance.
- Published Papers: DOI/link to published work. You must be listed as author. Journal name and publication date required. We verify publication independently.
- Teaching: Copy of slides/workshop materials, list of attendees or attendance certificates, learning objectives, and date/time of session.
- Courses: Course transcript or completion certificate with date and hours. Unofficial transcripts are acceptable if from an institution you're enrolled in.
- Standards: Documentation of your role (committee member, working group participant), your contributions, and dates of involvement. Email from standards organization confirming participation suffices.
- Case Studies: Link to published case study (on your company site, blog, publishing platform). Must be publicly accessible. We verify that you're listed as author/contributor.
- Mentoring: Names of mentees, their certification level (if earned), dates of mentoring relationship, description of mentoring activities.
- Open Source: GitHub/GitLab profile and specific PRs/issues showing your contributions. Your username and contribution dates must be verifiable.
Submit evidence through the EHC portal in your renewal application. You don't need to submit everything at once; you can accrue activities throughout your cycle and submit at renewal time. But don't wait until the last month; that's when authentication can take time and issues arise.
The Verification Process
When you submit EHC claims, eval.qa reviewers verify evidence. This ranges from quick checks (looking up a conference on the approved list and checking your name on the attendee list) to deeper review (reading your published paper to assess whether it's genuinely on evaluation topics, checking open source contributions on GitHub).
Most claims are approved within 2 weeks. Complex claims (standards contributions, mentoring arrangements that are less clear-cut) may take longer. If something is unclear, we ask for clarification. We operate from a position of good faith: you're documenting genuine professional activities. But we do verify.
Rejected claims are rare. They typically involve activities that don't meet criteria (internal company training with no external learning objectives, tool tutorials that are too basic, short talks that don't meet our documentation requirements). We explain the reasoning and suggest how to generate valid EHCs going forward.
Fraudulent EHC claims are treated seriously. Falsifying evidence or inflating participation violates the code of conduct, can result in credential revocation, and in extreme cases trigger professional review.
Fast-Track Recertification Options
Some certified evaluators don't want to pursue traditional continuing education. We offer alternatives that credibly demonstrate current competency without the standard EHC path.
Challenge Exam Option
Take a recertification challenge exam equivalent to the initial certification exam but with updated content reflecting field developments since your certification. You must score in the passing range to recertify. This is psychometrically equivalent to the initial exam; recertifying through this route proves you've kept current even if not through formal activities.
When: Available 6 months before your recertification deadline. You have until 3 days before deadline to schedule. Cost: 1/2 the cost of an initial certification exam. Why do this? If you prefer concentrated studying to distributed learning. If you weren't earning EHCs but want to recertify. If you want to validate your competency through assessment rather than activity tracking.
Roughly 10-12% of recertifications use the challenge exam route. Most are practitioners who've been in roles where they haven't accumulated documentation, or who prefer the psychological certainty of an exam over activity tracking.
Portfolio Recertification Option
Compile a portfolio demonstrating your work and current knowledge in evaluation. This isn't a resume; it's substantive work samples. Include: 2-3 evaluation projects you've led (with project summary, methodology, outcomes), analysis of 1 recent paper in evaluation/fairness/AI safety (2-page summary showing your deep engagement), 1 narrative essay on an evaluation challenge you've faced and how you addressed it (1,500-2,000 words).
Submit the portfolio for review by a panel of L4-L5 evaluators. They assess whether the work demonstrates current, advanced evaluation competency. This route takes more effort than standard EHC tracking but less than formal coursework. It's appropriate for practitioners who are deeply engaged in evaluation work but haven't documented activities for EHC purposes.
Portfolio review takes 4-6 weeks. If approved, you recertify. If additional evidence is needed, the panel gives feedback. If denied, you have the grace period to pursue standard EHC renewal or take the challenge exam. Roughly 5-8% of recertifications use portfolio recertification.
Competency Demonstration Option
For L4-L5 only. Present evidence of significant professional contribution to AI evaluation in the past 2-3 years. This might be: a major evaluation infrastructure or framework you built, published research advancing evaluation methodology, leadership of industry standards development, executive responsibility for evaluation at a major organization, creation of novel evaluation benchmarks or datasets.
This isn't the standard path (maybe 2-3% of recertifications), and the bar is high: you must demonstrate contribution of real significance to the field. But it acknowledges that some practitioners advance the field in ways that don't fit traditional EHC categories. The evaluation committee reviews your case and either approves recertification (because your contributions clearly demonstrate advanced current competency) or requests additional documentation.
What Happens if You Let Your Certification Lapse
You missed the grace period. Your L3 certification is no longer active. What now?
The First Year: Reinstatement
In the first year after lapse, reinstatement is straightforward. Show that you've earned the EHCs you would have needed during the lapsed period (or take the challenge exam, or submit a portfolio). Pay a small reinstatement fee ($50-100, depending on level). Your credential is active again with no loss of status.
This acknowledges that lapses happen: job transitions, family circumstances, sabbaticals. If you want back in quickly and can demonstrate you've kept reasonably current, the door is open.
After One Year: Reinstatement Exam Required
Beyond one year lapsed, you must take a challenge exam to recertify. This makes sense: you've been out of the credentialing system for over a year. Even if you've been working in the field, the field has moved on. An exam proves you're current enough to reclaim the credential.
This exam is full-price, same as an initial certification exam. The EHCs from the lapsed period don't carry over—you're starting fresh. This is a deliberate policy: we want certified practitioners to be actively engaged. A 3-year lapse followed by passing an exam and no continuing education going forward doesn't create the engagement we're looking for.
Distinction: Lapsed vs. Revoked
Lapsed credentials are temporary inactive status due to failure to meet CE requirements. Revoked credentials are permanently invalid due to violations of the code of conduct or professional standards. The distinction matters: you can reinstate a lapsed credential, but a revoked credential cannot be reclaimed.
Revocation is rare and requires serious violations: fraudulent EHC claims, proven incompetence in a high-stakes evaluation role, violation of confidentiality, harassment or discrimination, or gross negligence that causes harm. A single mistake or bad evaluation project doesn't revoke your credential. Revocation is for systematic failure to meet basic professional standards.
Building Continuous Learning Into Your Practice
EHC requirements are a minimum, not a maximum. Most practitioners earning their EHCs treat them as baseline and engage in deeper learning. How do you build a culture of continuous learning around your evaluation work?
The Learning Calendar Approach
Start each year by sketching a learning plan. What's changing in the field that matters for your work? What gaps do you want to close? What excites you? Map this to concrete activities:
- Q1: Complete one online course on emerging topic (AI safety, causal evaluation, etc.) — 12-15 EHCs
- Q2: Attend major spring conference — 15-20 EHCs
- Q3: Mentor junior evaluator, read and discuss papers with evaluation cohort — 5-8 EHCs
- Q4: Write case study on completed project, contribute to open source — 10-15 EHCs
This isn't rigid. Opportunities emerge (a workshop you didn't anticipate, a conference invitation, a research collaboration). But having a rough plan prevents November panic: "Oh no, I haven't earned EHCs yet." You're not trying to game the system; you're actively developing competency.
Learning Community Engagement
The most powerful professional development happens in community. Join or start an evaluation reading group. Find 3-4 other evaluators at your level or adjacent levels (L3-L4) and commit to monthly discussions of recent papers, tool releases, or case studies. This isn't formal EHC activity, but it deepens your understanding of the landscape and keeps you aware of developments.
Engage with online communities: eval.qa's member forums, AI safety Slack channels, fairness and accountability research groups, industry standards committees. Watching others' problems and solutions, contributing your own insights, staying in the discourse—this is where real learning happens. Much of it doesn't generate EHCs, but it's invaluable.
Teaching as Learning
One of the highest-value learning activities is teaching. When you teach others, you deepen your own understanding, discover gaps in your knowledge (students ask hard questions), and stay engaged with foundational concepts while pushing toward advanced ones.
This can be formal (teaching a university course, running a workshop) or informal (mentoring junior evaluators, leading internal team training). Both generate EHCs if structured appropriately. And both improve your own evaluation practice immeasurably.
Document your learning as you go. Keep a simple spreadsheet of activities, dates, and hours. When recertification comes, you're not scrambling to reconstruct what you did six months ago. You have a clear record. This also helps you see patterns: Are you over-relying on one type of activity? Are you staying balanced across learning formats?
Employer-Sponsored Continuing Education
Many practitioners receive employer support for professional development. This can significantly reduce the friction of meeting EHC requirements.
Optimal Employer Policies
Organizations that take professional development seriously typically offer:
- Learning budgets: $2,000-5,000 per employee annually for courses, conferences, books, and memberships.
- Paid learning time: 20-40 hours annually for professional development (reading, courses, attending workshops) as part of work time.
- Conference attendance: Coverage of 1-2 major industry conferences annually, including travel and registration costs.
- Internal learning programs: Regular lunch-and-learns, skill-sharing sessions, reading groups where teams discuss papers or tools.
- Tuition reimbursement: Coverage of formal courses (university courses, professional certifications) related to evaluation and AI safety.
When employers offer these programs, practitioners easily exceed EHC requirements. A full-time evaluator who gets 20-40 hours of learning time plus conference support can accumulate 40+ EHCs annually—exceeding even L5 requirements.
The Business Case for Supporting CE
From an employer perspective, supporting CE is investment in your team's capability. Evaluators who engage with continuing education stay current with field practices, catch issues others miss, build networks that improve your organization's reputation, and demonstrate engagement that improves retention. For organizations serious about AI evaluation quality, supporting CE is not cost—it's competitive advantage.
Some organizations negotiate directly with eval.qa: "We want our team to be well-certified and current. What CE investments would be most valuable?" We can advise on training priorities, help design internal programs, and work with organizations on scalable approaches to building evaluation capability.
Future EHC Categories
The EHC system is intentionally evolving. As the field develops, we add new activities that advance the field's goals. Here's what's under discussion for near-term addition:
AI Safety and Alignment Research
As AI safety becomes central to evaluation, we're creating formal EHC categories for safety research. This includes: formal participation in safety research programs (10-20 EHCs), collaboration on interpretability and mechanistic understanding research related to evaluation (8-15 EHCs), and contributions to AI alignment benchmarks and datasets (10-20 EHCs). These aren't approved yet but are in pilot programs.
Regulatory and Policy Engagement
The EU AI Act, NIST AI RMF, and emerging international standards require domain expertise. We're creating EHC categories for: regulatory consultation (5-10 EHCs), participation in policy development processes (8-15 EHCs), and evidence submissions to regulatory bodies (5-8 EHCs). This bridges professional practice and policy influence.
Synthetic Data and Simulation Expertise
As synthetic evaluation data becomes more important, we're considering EHCs for: developing synthetic evaluation datasets (10-20 EHCs), publishing work on synthetic-to-real transfer in evaluation (10-15 EHCs), and tools/frameworks for synthetic evaluation (8-15 EHCs). This recognizes that generating realistic evaluation scenarios is skilled work.
Equity and Inclusion Expertise
Fairness in evaluation continues to be complex and evolving. We're creating categories for: formal training in demographic analysis and fairness measurement (5-15 EHCs), community-engaged research with affected populations (10-20 EHCs), and publication on fairness in evaluation (10-15 EHCs).
EHC Recertification Checklist
- Know Your Deadline: Check the EHC portal. Know your recertification date and EHC requirement. Set calendar reminders at 6, 3, and 1 months before deadline.
- Plan Activities: Look at approved activities. Which fit your work and interests? Conference attendance? Publishing? Teaching? Mentoring? Build a rough learning calendar.
- Document Continuously: Don't wait until recertification. Keep a log of activities, dates, and hours as they happen. Saves scrambling at the deadline.
- Gather Evidence Early: Conference certificates, course transcripts, publication links—save these as you go. You'll need them for renewal.
- Submit Complete Applications: Incomplete applications delay the renewal process. Have all evidence ready before submitting.
- Use Grace Period Wisely: If you miss the deadline, use the 3-month grace period to get your affairs in order. Don't let the grace period expire.
- Explore Alternative Routes: If you didn't accumulate EHCs, challenge exam and portfolio options are available for most levels.
Stay Current With Your Certification
Join our community of lifelong learners in AI evaluation. Access the member portal, participate in our learning programs, and connect with evaluators at your level and beyond.
Explore Member Benefits