The Client
The Department of Computer Science at a federal university in Enugu had a problem that many Nigerian public institutions share: a legitimate need for modern computing infrastructure and a procurement budget approved two years earlier at prices that no longer reflected reality. The department head, Professor Ngozi Okafor, had been allocated funding for 20 student workstations as part of a lab upgrade initiative. The allocation was ₦28 million for hardware, installation, and a three-year maintenance agreement.
When she approached the standard government-approved IT vendors, the quotes came back between ₦38 million and ₦52 million for branded machines that she described as "not significantly better than what we already have." She reached out to Sephora Systems through a faculty colleague who had used us for a personal workstation build.
The Challenge
The lab's use case was specific: undergraduate computer science education. The machines would need to run:
- Programming environments: Visual Studio Code, Python, Java IDEs, GCC toolchains
- Database tools: MySQL Workbench, PostgreSQL, MongoDB
- Simulation and data science: MATLAB, Jupyter notebooks, R Studio
- Virtualisation: VirtualBox for OS and networking coursework
- Occasional: Blender (for a digital art elective), Unity (for a game development course)
The machines would be used by students who are not always careful with hardware. They would run in a room that is air-conditioned by a single 3HP split unit of uncertain reliability. They needed to be maintainable by the department's sole IT technician, who is competent but not a specialist. And they needed to last at least five years under daily institutional use — something branded consumer hardware rarely achieves.
Professor Okafor was also clear about one thing: she had been through a previous lab procurement where machines arrived that did not match the quoted specs. She wanted a vendor who would provide a detailed spec sheet before order and allow inspection on delivery.
The Consultation
We visited the lab space and did a thorough assessment. The room had one power circuit for the entire 20-station lab — a significant limitation. Running 20 PCs simultaneously on a single circuit meant we needed to design for low power draw without sacrificing performance. We also noted that the existing air conditioning was barely adequate for the space; we recommended the department requisition a second unit before installation, which Professor Okafor pursued through a separate budget line.
We proposed a build optimised for the academic workload: strong multicore CPU performance for compilation and virtualisation, adequate GPU for the occasional 3D coursework, and prioritised storage performance for database and data science work. We would also provide a documented build specification, delivery manifest, and a three-year parts warranty on every machine — something the branded vendors had not offered at comparable prices.
The Build
20 Academic Workstations — ₦1.18 million each (₦23.6 million total):
- CPU: AMD Ryzen 7 7700 — 8 cores, 65W TDP (important for the power circuit constraint), strong single-core for compilation
- RAM: 32GB DDR5 — sufficient for VirtualBox sessions and MATLAB without paging
- GPU: NVIDIA RTX 3050 8GB — handles Blender and Unity coursework; supports dual monitors
- Storage: 1TB NVMe (OS + software) + 1TB SATA SSD (student project storage)
- Case: Fractal Design Pop Air — tool-free access panels, dust filters, easy maintenance
- Cooling: DeepCool AK500 air cooler — silent, reliable, no liquid maintenance overhead
- UPS: APC 1000VA per machine — 20 units on a staggered power load to stay within circuit limits
Remaining budget (₦4.4 million) covered: a managed network switch, 20 24" 1080p monitors, structured cabling, a shared 40TB NAS for student project backup, and our three-year maintenance agreement including annual cleaning and thermal paste replacement.
Total within budget: ₦28 million even. We provided a full itemised spec sheet with component serial numbers before order confirmation. On delivery day, Professor Okafor's IT technician inspected five randomly selected machines against the spec sheet. All matched exactly.
The Result
The lab opened for the 2026/2027 academic session. Professor Okafor reported that compilation times for student Java and C++ projects — a proxy for general responsiveness — were dramatically faster than the previous lab's machines. MATLAB simulations that previously took 8–12 minutes complete in under 2 minutes. VirtualBox sessions that previously caused machines to swap and stall are stable with the 32GB RAM headroom.
The IT technician's first maintenance report, three months in, noted that dust accumulation in the Fractal cases was significantly lower than in the department's older machines — a direct result of the magnetic dust filters and positive-pressure airflow design. He estimated he was spending 60% less time on reactive maintenance than with the previous lab.
Key Takeaway
University and institutional procurement in Nigeria is constrained by budget cycles and approval processes that often mean the money available doesn't match current hardware prices from standard vendors. Custom building — with transparent specifications and verifiable component choices — is frequently both cheaper and better than branded alternatives in this segment. The key is a vendor relationship built on documented specs and accountable delivery. If a vendor can't show you exactly what they're building before you pay, that's your answer.
Running an educational institution or research facility that needs computing infrastructure? Talk to our team — we work with universities, schools, and research labs on spec-to-installation projects.