Skip to content

98% faster page loads and a 12% smaller cloud bill for a UK careers charity

A UK charity whose platform connects students and graduates with 600+ professionals sharing career insights across 80+ career paths.

98%
Faster page loads
Measured at go-live, week 8
12%
Lower cloud cost
Against the previous AWS bill, on migration to Hexploits Cloud
2 days
To ship a change
From 3+ weeks, measured after go-live
8 weeks
From first day to production
Nov 2024 – Jan 2025

In brief

  • Hexploits rebuilt Gradvisor’s careers platform in eight weeks after the previous supplier’s build reached ten-second page loads and three-week change cycles.
  • Page loads fell from 10 seconds to 150 milliseconds (98% faster) and time to ship a change fell from three weeks to two days.
  • Migration to Hexploits Cloud cut the hosting bill by 12% against the previous AWS bill and fixed it monthly.
  • The platform moved from manual, untested releases to automated, tested releases, and has been under Hexploits managed support since.

Client context

Gradvisor is a UK charity. Its platform connects students and recent graduates with more than 600 professionals sharing first-hand career insights across 80+ career paths. For a young person without a network, it is the network.

The charity runs on grants and donations, so every pound spent on technology that does not work is a pound taken from the mission. It has a small staff and no in-house engineering team.

The situation before

The platform had been built quickly by a previous supplier and had reached the point where it was working against the charity rather than for it.

  • Pages took around ten seconds to load. Students left before the content appeared.
  • A minor change took three weeks or more to reach users. The roadmap had stopped.
  • Infrastructure spend could not be forecast, which for a grant-funded organisation is an operational risk rather than an inconvenience.
  • There were no automated tests and releases were manual. Every deployment was a gamble, and regressions were the norm.

Underneath, the data had been put in a document database although it was relational, so every feature fought the storage layer. The previous team had moved on.

What we did

We recommended a rebuild rather than a patch, because a patch would have cost more over the life of the platform, and we said so in writing before the work started.

Over eight weeks we redesigned the system, migrated the data with no loss, rebuilt the platform on mainstream technology any engineer can maintain, and introduced the practices that make releases routine: automated tests, environments that match production, and a pipeline that checks every change before it reaches a student.

We rebuilt the features that make Gradvisor more than a library: a recommendation engine that matches students to career paths and people, an assistant that answers questions from the charity's own 6,000+ insights, and a voice pipeline that turns recorded interviews into searchable text. All of it was built to cost in proportion to use.

After go-live we migrated hosting to Hexploits Cloud, added reporting and monitoring, and moved the relationship onto a managed support contract. We have run the platform since.

Results

MeasureBeforeAfterPeriod
Page load time10 seconds150 millisecondsMeasured at go-live, week 8
Time to ship a change3+ weeks2 daysMeasured after go-live
Release processManual, untestedAutomated, tested on every changeFrom week 6
Cloud costPrevious AWS bill12% lower, fixed monthlyOn migration to Hexploits Cloud
Voice recordings made searchable05,320 recordings, 68,945 transcriptionsAt time of writing

For a charity, the second row matters as much as the first. A platform that can change again means a roadmap the trustees can plan around.

What the client said

Fantastic company and our development partner for Gradvisor, a social mobility careers platform with national ambitions. Extremely responsive and mission-oriented. Cameron owns any shortfalls humbly - rare for IT providers. Thinks like a client too.

Tushar Prabhu, Founder, Gradvisor

For the IT reader: technical detail

Architecture. The original platform kept users, careers, recommendations and connections in DynamoDB. We moved 50+ tables to a normalised Postgres schema and redesigned 170 API endpoints with typed contracts. Data migration was scripted, rehearsed against a copy of production, and verified row-for-row before cutover.

Platform. Next.js and TypeScript for the application, Postgres for transactional data, infrastructure as code for every environment, and GitLab CI running unit and integration tests on every merge request. Development, staging and production match in shape.

Intelligent services. The recommendation engine scores students against career paths, roles, sectors and other members on stated interests, engagement and skills. The career assistant uses retrieval-augmented generation over the user's profile, the full content library and live opportunities. The voice pipeline transcribes recordings, validates quality and indexes the output for retrieval.

Hosting and operations. After stabilisation the platform moved from the client's AWS account to Hexploits Cloud: the same architecture on reserved capacity with faster CPUs and more memory, at a fixed monthly figure 12% below the previous bill. We run deployments, patching, backups and incident response under the managed support contract.

Reporting and observability. Metabase for self-service reporting to staff and funders, PostHog for product analytics and feature flags, Datadog for infrastructure monitoring, alerting, logs and traces.

Compliance. GDPR controls built into the platform (consent management, erasure and portability workflows, data minimisation) and security practices aligned to ISO 27001: least-privilege access, incident procedures, audit trails and regular assessment. Alignment, not certification; see the trust page for our own accreditations.

Questions about this engagement

Why rebuild rather than patch?
The data had been put in a document database although it was relational, and there were no tests. Patching would have cost more over the platform’s life; the recommendation and the reasoning were put in writing before work started.
How was the migration done without losing data?
Scripted, rehearsed against a copy of production and verified row-for-row before cut-over; 50+ tables moved from DynamoDB to Postgres and 170 endpoints were redesigned.
What does Hexploits still do for Gradvisor?
Hosting on Hexploits Cloud, monitoring, patching, backups, incident response and small changes under a managed support contract.

Have a system like this one?

Tell us about the system and the sector. A named engineer replies within one working day. A written scope and an indicative price within two working days of a short scoping call.