Summary

  • LACNIC and its partners built Ayitic around several steps between training and economic participation. Public evidence records Swift's Caribbean coordination and project responsibility, while credits assign design, local delivery and evaluation to a wider group.
  • The record supports a distinction between enrollment, graduation, job applications, platform customers, internships and confirmed jobs. It does not reconcile two sets of program totals or establish a single employment rate for all participants.

The first question after a digital-skills course is not whether a student can perform a lesson. It is what happens when she tries to turn that lesson into paid work. In the public account of Ayitic, that question appeared before the program's results could be reduced to a graduation number. At a 2018 Internet Governance Forum session, participants discussed women who had completed training and were beginning to apply for online jobs. The same record says the conversation turned to the difficult transition from training to actual work. The gap was visible in the program's own early description: a course could end on schedule while the labor-market test was only beginning. (IGF session record)

That is the useful place to begin an account of Kevon Swift's connection to Ayitic. It keeps the subject's documented role in view without making him stand in for every organization that designed, delivered, assessed or funded the program. It also keeps the program's stated ambition separate from the results its public evidence can establish. Ayitic was not simply a set of classes.

It was an attempt to build a bridge from skills to a market in which employers and clients needed reasons to trust a new worker, workers needed the means to receive payment, and access to electricity, devices, time and connectivity affected whether a lesson could be used outside the classroom.

The record places Swift in coordination, not sole ownership

LACNIC's November 2014 interview described Swift as its newly appointed External Relations Officer for the Caribbean. It presented his job in terms of regional engagement and relationships with Caribbean stakeholders. The interview also described Ayitic as a strategic intervention connected to LACNIC's wider work. The following year, an Internet Society biography said Swift had joined LACNIC in August 2014, led Caribbean engagement, coordinated technical training and had recently been assigned responsibility for Ayitic, described as the organization's flagship Internet-for-Development project in Haiti. Those are meaningful responsibilities. They do not, by themselves, identify him as the sole author of the program's curriculum or delivery model. (LACNIC's 2014 interview; Internet Society biography)

The distinction matters because Ayitic depended on several kinds of work that sit outside a regional-relations title. LACNIC's 2016 account of the initiative says Development Projects Coordinator Carolina Caeiro had emphasized local context, local instructors and the responsibilities of partner organizations. The 3x3 Design final assessment later credits Caeiro and Beatriz Irisarri at LACNIC for program and curriculum design, Caribbean Open Institute for curriculum design, École Supérieure d'Infotronique d'Haïti (ESIH) for course delivery, and 3x3 Design for the evaluation. A record that says Swift attended the opening or was assigned project responsibility is not a substitute for those named credits. (LACNIC's 2016 Haiti account; 3x3 Design final assessment)

The evidence therefore supports a narrower and more useful description: Swift was part of the leadership and coordination surface through which LACNIC engaged Caribbean stakeholders and took Ayitic forward. The public sources do not assign him every operational task. Nor do they establish that every outcome should be read as either his success or his failure. A multi-organization project distributes responsibility by design. Person-level reporting has to preserve that structure rather than compress it into a biographical hero story.

This is the same discipline that applies to participation in governance. Attendance, responsibility, authority, implementation and measurable outcome are separate propositions. A person can be present at an opening without designing the intervention; a person can be responsible for coordination without controlling every local delivery condition; a participant can complete a class without controlling the platforms, payment rails or household constraints that determine whether the skill becomes income. The distinction does not diminish work. It makes the attribution accurate enough to be useful.

Ayitic was a pathway, not one training count

The project was designed to address several related but different constraints. The International Development Research Centre describes project 108360 as a completed, 30-month, CAD 1.147 million partnership led by LACNIC and ending on 11 March 2020. Its stated objective was to help young Haitians find work in the digital economy through locally adapted courses, matching platforms and training for telecommunications engineers. The final assessment describes three strands: digital skills and work pathways for women; network-management and security training; and work with partners to scale or extend the model. These are connected activities, but they do not share a single employment denominator. (IDRC project record; 3x3 Design final assessment)

A participant learning to sell digital services online faced a different route to work from a technician studying network management and security. The first route depended on a client understanding a profile, trusting a portfolio, selecting a worker through a platform and transferring money across a payment system. The second required technical instruction, equipment, laboratory time, employer demand and a way to demonstrate competence in an infrastructure role. A program could make progress in either route without producing the same form of employment evidence.

Counting everyone as though they had passed through one identical pipeline would make the total look cleaner while reducing its meaning.

The chronology also has more than one frame. IDRC's project record gives the funder's 30-month project period. LACNIC's annual reporting describes a longer, three-year story of Ayitic's evolution. The 3x3 assessment examines three training rounds delivered between March 2018 and August 2019 and refers to the evaluated initiative as a two-year partnership. Each description can be correct within its own reporting scope. None should be silently used to overwrite the others. The difference is a reminder that an initiative's funding clock, public program history and evaluation window answer different questions.

Ayitic's public communications also emphasize different moments. In 2017, LACNIC highlighted the project at the G20 summit, placing it in a wider conversation about digital development. In 2018, it described the launch of three courses in Haiti and the completion of an initial women's cohort. In 2019, the annual report summarized participant reach and reported work-related figures. The final evaluation, published in the same broad period, considered participant outcomes and implementation. These communications are not interchangeable measurement instruments: an announcement explains a project's purpose; an annual report supplies organizational figures; an evaluation examines performance through a particular design. (LACNIC's 2017 account; LACNIC's 2018 course launch; LACNIC 2019 annual report)

Graduation was an output; paid work was another test

LACNIC's account of the first women's cohort says 50 participants completed a 13-week course. It describes tablets, an online learning platform and support for employment beginning after course completion. That is evidence of a defined cohort reaching a training milestone and of an intended next step. It is not evidence that every graduate entered paid work, that work began immediately or that one intervention caused a later contract. A certificate or completion record answers an education question. A paid engagement needs a different record: a client or employer, a date, a defined task, a payment or compensation status, and a way to distinguish completed work from an application. (LACNIC's first-cohort account)

The gap is not a semantic technicality. Online work marketplaces can make access appear frictionless because registration is digital and opportunities can be browsed from a phone or computer. But a marketplace does not remove the costs of proving skill. A new worker may have to prepare a portfolio, write a proposal, bid against established sellers, supply reviews or references, absorb time spent on unsuccessful bids and wait for a client to trust the profile. If a platform ranks reputation, an entrant with no work history is not competing on the same informational footing as an experienced seller.

If applications consume paid or unpaid time, a graduate who cannot afford to keep bidding may leave before the market recognizes her ability.

The final 3x3 evaluation treats training completion and employment as separate results. It reports that efforts to connect graduates with jobs in the domestic market and through online platforms had limited success. The evaluators point to practical barriers: lack of experience and reviews, missing referrals, credentials, payment access and language capability. They also describe the effort and time required to bid in online marketplaces. This evidence does not say that training had no value. It says that skill acquisition was only one input in a chain where the person seeking work also needed credibility, market access, transaction capability and enough time to persist. (3x3 Design final assessment)

A training-to-work chain can be written as a sequence: enrollment, attendance, learning, completion, proof of skill, a credible introduction, application, selection, contract, delivery, payment and repeat work. Each transition can fail for a different reason. A course may be well taught while the graduate lacks a certificate accepted by an employer. A certificate may be accepted while there is no referral. A referral may lead to an interview, but a job may require reliable power or schedule flexibility. A task may be won, but cross-border payment may be unavailable or too expensive.

The project evaluation's value is that it places these frictions in view. Its limits are that it does not estimate the separate causal contribution of each barrier for the whole participant population.

The numbers have to stay attached to their sources

The 3x3 overview reports 316 women in the Internet and Data Practitioner (IDP) track and 169 trainees in Network Management and Security (NM&S), or 485 participants across the two tracks. Its detailed tables give a different NM&S count of 166. The report's detailed IDP section lists 316 registered participants, nine dropouts and 257 graduates. For NM&S it lists 166 trainees, 11 dropouts and 81 graduates. Its summary describes graduation and attrition rates of approximately 84% and 3% for IDP, and 52% and 7% for NM&S. These figures should be represented as the evaluation's own reported figures, not recalculated into a new combined rate. (3x3 Design final assessment)

LACNIC's 2019 annual report gives a separate set of numbers. It reports that Ayitic had trained 358 women, with 86% graduation and 3% attrition in the IDP component. For NM&S it gives 163 technicians trained, 59% graduation and 14% attrition. It also says six internships had been completed, two jobs had been confirmed and two customers had been obtained through Freelancer. The annual report states that courses later reached more than 3,000 students on LACNIC Campus. These are valuable organizational disclosures, but the available material does not explain how their counts and denominators relate to those in the 3x3 report. (LACNIC 2019 annual report)

The disciplined conclusion is not that one report must be wrong. It is that the reporting scopes are not reconciled in the public record reviewed here. The difference might reflect timing, cohort boundaries, updated records, definitions or another change, but those are possibilities rather than established explanations. It would be inaccurate to select the higher graduation rate because it sounds more successful, select the lower one to make a critical point, or average them. A source-attributed count is stronger than a harmonized number whose method no one can reproduce.

The same rule applies to work-related terms. Two Freelancer customers are not necessarily two individuals employed; a customer is a buyer-side measure. Six internships completed and two jobs confirmed are distinct categories, and the report does not say whether they overlap with the customer count or the evaluation's respondent group. Job applications are a leading activity, not a placement. A completed course is a training output, not a placement. A later learner in LACNIC Campus may show reach, not an Ayitic employment outcome. Keeping these labels intact is essential to a useful public account.

The 3x3 assessment uses multiple methods. Its overview describes 450 trainee survey responses, eight focus groups and 22 individual interviews, and says the evaluation also spoke with the employment coordinator and employers. Its post-completion impact surveys are smaller: N=119 for IDP and N=65 for NM&S. A sample can offer important evidence about what participants encountered, but a respondent-level barrier should not be transformed into a cohort-wide rate unless the report provides a design and denominator that support it. A focus group can explain a mechanism; it cannot tell us how many of all trainees experienced it.

An employer interview can reveal a hiring constraint; it is not a market census.

These limits are not reasons to dismiss the evaluation. They are reasons to ask the right question of each evidence type. Program records can establish enrollment and completion under a defined administrative rule. Surveys can report what respondents said. Interviews can show how a process felt and where it broke. Employer accounts can illuminate hiring expectations. A job record can verify a placement under a specified definition. None automatically substitutes for the next.

If a later evaluation were to align its cohort IDs, time windows, definitions of work and payment status, it could explain what the existing annual-report and evaluation numbers cannot presently explain.

The friction sits beyond the classroom

The online-marketplace barriers in the evaluation are not merely about software. A platform profile is a bundle of signals: prior jobs, ratings, reviews, referrals, credentials, portfolio examples, responsiveness and language. A graduate can have technical ability but little proof that a remote buyer can verify. The platform may appear open to anyone, while the route to trust is path-dependent. A worker who needs the first rating must win an initial task; a buyer who wants low risk may choose a worker with a record. The course can teach how to use a platform, but it cannot by itself produce a market's accumulated trust signals.

This is why the assessment's recommendations are operational rather than rhetorical. It suggests practice in bidding and negotiation, training in payment systems, credentials available soon after graduation, referrals or testimonials, alumni connections and outreach to employers before a student finishes. Each recommendation addresses one point where the transition can stall. Bidding practice can reduce avoidable mistakes, but it cannot guarantee a client. A certificate can make skill more legible, but it cannot replace every employer's experience requirement. A referral can lower the cost of trust, but the referral network must connect to actual demand. The evaluation names possible mechanisms; it does not establish that the suggested measures were adopted or that they would have produced a specific employment total. (3x3 Design final assessment)

Payment access is especially easy to understate. A worker can find an online task, satisfy a client and still face limits on how money is received, held or converted. The final assessment lists payment access as one of the barriers participants encountered. That makes the payment rail part of the employment pathway, not an administrative detail after success. Training in a service skill may raise a person's ability to deliver work; it does not alone establish access to the financial and platform systems through which work is sold and paid.

The project's stated use of matching platforms and the evaluation's payment findings therefore describe opposite ends of one chain.

Language also interacts with reputation. A worker may have relevant knowledge but be unable to write proposals in the language buyers expect, explain scope, negotiate revision terms or handle disputes. If a task requires client communication, language proficiency is not a generic cultural attribute; it is an operating capability tied to a particular market. The evaluation names language as one factor, alongside experience, referrals and payment access. The sources do not quantify the standalone effect of each, so it would overstate the evidence to rank language above another barrier or make it a complete explanation.

Domestic work and international platform work create different alternatives. The assessment says the program had limited success connecting graduates to both domestic opportunities and online platforms. That distinction matters for program design. A domestic employer may value a different credential, schedule, language or connection than a remote marketplace buyer. A platform may expose someone to a wider set of tasks yet also intensify bidding, rating and payment constraints. Treating “digital jobs” as one market hides the question of which demand existed, which route participants could reach and what evidence of work each route produced.

Infrastructure and time shape the feasible choice set

The assessment also records constraints that no short skills course could fix by itself: devices, electricity, Internet access, household and care responsibilities, time, and political instability. These conditions can interrupt learning, limit job searching, make a promised schedule unreliable or raise the cost of delivery. They are not evidence that participants lacked talent or that a course was poorly designed. They are part of the environment in which any training-to-work intervention has to operate.

A connectivity-dependent work program can be evaluated as if access were a binary variable: the person is online or offline. Daily practice is less binary. A connection can be intermittent, data can be costly, a device can be shared, electricity can fail during a timed application, or household duties can determine when a person can log in. A platform bid may require a continuous block of attention and a response within a narrow window. The public reports do not provide enough detail to quantify the effect of each condition, but their presence changes how the employment barrier should be understood.

The skill does not travel into the labor market in isolation; it travels through devices, networks, payment services and available time.

The same is true of care work. If a participant carries responsibilities at home, an unpaid bid or an uncertain client may be a more expensive experiment for her than for another person with more discretionary time. A program can provide access to training and still leave the costs of proving oneself in a marketplace with the individual. That is not automatically a defect in the curriculum. It is a distributional fact about who absorbs the risk while waiting for a first contract. The evaluation points toward that burden without offering a single numerical estimate of it.

Political instability can affect both sides of the market. It may change whether people can attend, whether employers continue operating, whether networks and electricity remain reliable and whether a client can transact safely. The assessment lists instability among the barriers but does not isolate a causal coefficient. A careful article should not turn a broad contextual factor into a convenient explanation for every missing job, or imply that it affected every participant in the same way. It belongs among the constraints that shape what a training program can plausibly promise and measure.

This is where regional coordination becomes relevant, but not as a substitute for local evidence. A regional organization can help assemble partners, training methods and technical expertise. Local organizations can understand teaching conditions and relationships to prospective workers. Employers can report what hiring actually requires. Participants can describe the costs and tradeoffs they face. A credible pathway must connect those perspectives without confusing who supplied which information. The available record documents some of that division of work, while leaving several outcome definitions and denominators unresolved.

Local delivery changes what coordination can accomplish

The 2016 LACNIC account describes a shift toward training local instructors and working with organizations that understood Haitian context. The 3x3 credits make the distribution of tasks more concrete: LACNIC staff are named for program and curriculum design, Caribbean Open Institute for curriculum design, ESIH for course delivery and 3x3 for assessment. This is more than a list of partners. It shows that “the program” was an assembly of design, instruction, local adaptation, employment support and evaluation, each with a different vantage point and accountability surface.

Local delivery is often described as an advantage, and it can be one. A local partner may know which examples are usable, which devices learners can access, how a training schedule intersects with daily life and which employers or institutions can be approached. But local presence does not automatically solve demand, payment or trust problems. It makes those conditions more visible and gives a project a route to respond. The difference between understanding a barrier and removing it should remain explicit.

The public report's staffing and credit record also limits what can be said about Swift. The Internet Society biography records a project-responsibility assignment; LACNIC's account records his attendance at the 2016 opening; and the IGF listing later names him as an organizer and rapporteur/notetaker. Those entries show sustained connection to the initiative's public and coordination surface. They do not say he taught the courses, selected each participant, managed each local partner, negotiated all placements or controlled Haiti's employment and infrastructure environment.

No source in this package supports those broader attributions.

That boundary protects both the subject and the story. Assigning a collective outcome to one visible person can make an article seem decisive while making the underlying organization less legible. It can credit a person for work named sources attribute to colleagues or local institutions, or blame one coordinator for constraints outside the role described in the record. The better biographical question is what kind of work his role made possible and what the record says about the limits of that role.

In this case, it connects Caribbean engagement and project responsibility to an effort whose delivery and assessment depended on a wider set of actors.

A participant story can illuminate a mechanism, not a rate

LACNIC's 2019 interview with Waina Moise gives the program an individual voice. Moise describes discovering digital tools, using job platforms and working with the employment coordinator. Her account can show what one graduate tried and how support looked from her perspective. It should not be made to carry a claim about what happened to all graduates. One person can demonstrate that a pathway was possible for that person under particular conditions; she cannot establish the probability of success across a cohort. (LACNIC's interview with Waina Moise)

The distinction works in the other direction too. A low count of confirmed jobs in an annual summary cannot tell us what every participant learned, whether they used the skills later, whether work happened outside the tracked period or whether a graduate chose another route. Employment measures need a defined window, a definition of paid work, a treatment of freelance and informal work, and a way to handle work found without program assistance. The 3x3 assessment concludes that connecting graduates to work had limited success. That is an important program-level finding.

It is not a verdict on every student or a complete measure of long-term value.

A person-centered article should resist turning a participant into a symbol. The Waina interview should not be used to prove that Ayitic succeeded because her story is compelling. Nor should missing aggregate placement data erase her account or make it irrelevant. The account is a qualitative piece of evidence about one route and one participant's experience; the survey and assessment supply broader but still bounded evidence. Taken together, they help readers see both a particular mechanism and the limits of the available denominator.

Scale requires a new question, not just a larger number

LACNIC's annual report says Ayitic courses later reached more than 3,000 students on LACNIC Campus. That is a notable reach figure. It suggests that some of the educational material continued beyond the initial Haitian cohorts and could be delivered at a larger scale. It does not establish how many of those learners were in Haiti, whether they belonged to the original program, completed a course, entered the same type of work pathway or earned income as a result. Reach is a different output from completion, and both are different from employment.

At scale, the measurement problem can become harder rather than easier. A platform can count registrations, views and certificates cheaply. A reliable record of job quality may require follow-up across employers, freelance clients, informal work and time. If the course is open across geographies, the context that made the Haitian intervention relevant may not travel intact. If employment support is attached only to a cohort, later self-directed learners may not have the same referrals or payment guidance.

An expanded course footprint therefore invites a new question: what elements of the original support model were replicated, and what evidence follows learners into work?

The annual report is valuable because it distinguishes, at least in its wording, some separate outputs: women trained, technicians trained, internships completed, jobs confirmed, Freelancer customers and later Campus reach. A reader should preserve those categories. A business customer count is not the same as an individual worker count; internships are not necessarily sustained jobs; and course reach is not an earnings outcome. The report's summary does not provide a common tracking framework that would let the reader infer how these outputs overlap.

There is no need to manufacture a verdict from incomplete data. A program can have meaningful educational value, build technical capacity and still leave employment pathways weaker than its designers hoped. Conversely, a limited count of confirmed jobs in a report does not prove that learning produced no long-term benefit. The evidence supports a bounded conclusion: Ayitic attempted to join training to work, reported some job-related activity, and encountered barriers at the point where credentials, trust, payment, time and demand had to meet.

The record does not deliver a single, reconciled placement rate or a causal estimate of the program's employment effect.

What Swift's public record can and cannot carry

Swift's documented career role helps explain why Ayitic sat within LACNIC's regional work. In 2014 and 2015 sources, he appears as a Caribbean engagement lead and a coordinator of technical training, with responsibility for Ayitic assigned within the organization. In 2016 he is recorded at the opening, and in 2018 he is an organizer and rapporteur for a session that discussed the distance between training and work. These are not trivial mentions. They place him in the institutional path connecting regional engagement, capacity building and public discussion of the project.

What they do not supply is a personal impact assessment. The sources do not isolate his decisions from those of the named program designers, local delivery organizations, employment coordinator, funders, employers or participants. They do not say which project decisions he made, which targets he controlled or how his own work affected the 3x3 evaluation's results. This absence should not be converted into an insinuation. Public biographies and reports have limits; the appropriate response is to state what they document and stop where they stop.

The point is particularly important in infrastructure institutions, where coordination can be consequential but is often difficult to observe. Convening a partner, arranging a training relationship or putting an initiative on an international agenda may influence what is possible. Yet influence is not the same as sole control, and presence is not proof of a causal result. A role can matter while the record remains insufficient to allocate outcome credit. That is not an argument that nothing happened. It is an argument for keeping the level of certainty matched to the source.

The same evidentiary discipline applies to the program. The evaluation's findings are not a referendum on Swift; the annual report's counts are not his personal scorecard; a participant's testimony is not an endorsement of a manager. Ayitic is best understood as a partnership with defined responsibilities and a difficult labor-market interface. Swift's public role is part of that structure, not an alternative to it.

The missing evidence is operationally specific

A stronger public record would not need a grand new narrative. It would need a small number of aligned fields. Each cohort could be linked to a stable participant count; course and graduation definitions could be identical between the annual report and evaluation; and follow-up could specify a date window and whether work meant full-time employment, an internship, a paid freelance contract, an unpaid trial, a platform customer or a self-reported opportunity. The report could also say whether counts overlap and whether a job was directly connected to program support.

Those definitions would make the existing figures easier to interpret without changing the underlying program.

The transition could be measured at several points rather than collapsed into one success/failure label: a graduate receives a verifiable credential; creates or improves a portfolio; submits a defined number of bids; is contacted by a buyer; completes paid work; receives payment; obtains a repeat customer; or secures an internship or employment contract. These indicators would not all be outcomes of equal value, and a bid count should not become the target if it rewards activity without income. But a sequence makes it possible to identify whether the bottleneck lies in skill, proof, market access, selection, delivery or payment.

The evaluation's recommendations point toward some of these steps: bidding and negotiation exercises, payment-system instruction, timely certificates, referrals, alumni networks and employer outreach before graduation. A future report could distinguish which of those were implemented, for whom, and with what observed result. If a credential improved recognition but no one completed paid work, that would be informative. If referrals increased first contracts but not repeat work, that would be informative too. The available sources do not document this follow-through, so the article cannot claim it happened.

There is also a question of how to handle context fairly. If access to electricity, devices or stable connectivity was interrupted, the evaluation could record the timing and intensity of interruptions without treating them as individual shortcomings. If care responsibilities affected participation, follow-up could distinguish schedule flexibility from course content. If platform payment rules blocked transactions, the program could record where a candidate reached in the payment chain. Those are not demands for intrusive data collection; they are examples of specifying the point where an intended pathway met a constraint.

A measurement design should also make room for outcomes that are valuable but not identical to employment. Technical confidence, a certificate, a new professional contact, an ability to continue learning, or a better understanding of online transactions can matter to a participant. They should be reported under their own names rather than recoded as jobs. Clear categories do not belittle non-employment value; they allow readers to understand it without mistaking it for an income result.

The larger lesson is about the boundary of a program

Ayitic's ambition can be understood as a response to a real boundary problem. Training providers can help a person learn; they cannot, by that fact alone, compel a buyer to trust her, guarantee a client exists, create a payment option or remove the local costs of being online. A program can influence some transitions by adding credentials, practice, introductions or employment support. It cannot make every external dependency disappear. A public account becomes more useful when it identifies which part of the chain the program controlled and which part remained outside it.

This is also why numbers need a provenance trail. “Trained” could describe enrollment, participation in one module, completion of a sequence or a count of certificates. “Employed” could include an internship, a freelance task, a customer relationship or a formal job. If sources use different scopes, each figure can be true under its own rule and still be incomparable. The public record should carry a label, source, time window and denominator beside each outcome. Without that, a number can travel farther than the evidence that gives it meaning.

For readers interested in Swift, that boundary prevents a familiar biographical distortion. A person associated with a program may become its narrative center, even when partners performed the design and delivery and when the actual market response depended on actors outside the project. The record here gives a genuine reason to include Swift: his regional role and assignment tied him to Ayitic's coordination. It gives no reason to turn him into either the sole architect of success or the owner of unresolved employment constraints.

For readers interested in development programs, the boundary changes the evaluation question. Instead of asking only how many people finished a course, ask what evidence carried each graduate into the next transaction. Was skill made legible? Could a prospective buyer verify it? Could the worker submit, negotiate, deliver and get paid? Were local employers part of the design before graduation? What was the cost in time and connectivity of reaching the market? Where did a participant stop, and what source confirms that point? Those questions do not promise a particular answer. They make a program's operating surface visible.

A careful conclusion leaves the uncertainty intact

The public record supports three conclusions. First, Ayitic was a multi-part effort to link digital and technical training with economic participation in Haiti, with LACNIC and local partners sharing different design, delivery and coordination roles. Second, the sources place Swift in Caribbean engagement, technical-training coordination and project responsibility, then record his participation in public program events. Third, the project reports indicate a gap between completing training and obtaining work, while the evaluation documents marketplace, payment, language, credential, infrastructure and time constraints.

The record also leaves material questions open. It does not reconcile the 3x3 and LACNIC annual-report counts. It does not provide one placement denominator across all tracks and cohorts, define exactly how the customer, internship and job figures overlap, or estimate a causal employment effect. It does not assign all operating decisions or outcomes to Swift. Those are not reasons to assume the worst or to celebrate an unproven result. They are the limits of what these sources can bear.

A useful portrait of a program leader can hold both parts at once: a documented role in building connections and an outcome chain that no single coordinator controls. Swift's record connects Caribbean engagement to a project trying to take training beyond the classroom. Ayitic's evaluation shows that the distance from a course to paid work was not a final administrative step. It was another system, with its own trust rules, transaction infrastructure, time costs and demand.

The clearest account is the one that describes that system without asking a graduation figure or an individual biography to stand in for evidence that is not there.

Sources