Engineering perspectives: enterprise AI case studies without the names
We publish engineering perspectives rather than conventional enterprise AI case studies, because our clients' names and numbers belong to them. Each write-up below describes a real engagement: the operating context, the design choices and what the work involved, labelled by industry and capability. Metrics and identities are withheld until the client approves them.
Selected engineering perspectives
Four engagements, written from the engineering record. Every published perspective is reviewed for accuracy, and client permission is requested before any name appears.
Digital onboarding platform for a securities firm
Industry: financial services, India. Capability: enterprise application development. A securities firm needed customers to open and verify accounts digitally, with the licensed integrations the regulated process requires. We built the onboarding platform and its integrations, with the document and identity steps designed for audit from the first release. Client name withheld pending permission.
Multi-environment GKE observability and load testing
Industry: technology platform. Capability: cloud engineering and DevOps. A platform running on Google Kubernetes Engine needed separate dev, test and production environments with one observability view. We designed a two-cluster topology with centralized Grafana, Prometheus and ArgoCD, and load-tested the platform at a 200,000 concurrent-user target before release. Client name withheld pending permission.
Infrastructure monitoring platform for GCC sites
Industry: energy and facilities, Gulf region. Capability: edge and monitoring engineering. Distributed sites with constrained connectivity needed one monitoring platform. We built a pipeline on Telegraf agents, EMQX for MQTT transport, TimescaleDB for time-series storage and Zabbix proxies at the edge, designed to buffer and resynchronize when links drop. Client name withheld pending permission.
LangGraph assistant for multi-property hospitality
Industry: hospitality. Capability: generative and agentic AI. A hotel operator wanted one assistant for room booking, room service, food ordering, feedback and checkout across properties. We built a multi-tenant agent on LangGraph with PostgreSQL checkpointing, Redis sessions and Indian-language speech, and benchmarked LLM against retrieval intent classification before choosing. Client name withheld pending permission.
How we write a perspective
Each write-up is built from the engineering record, not from marketing notes: the architecture decisions, the trade-offs we argued about and the operational outcome. We describe the client's type and sector but never the name, and we leave out metrics unless the client has approved the measurement period and the figure. When a perspective is approved for publication in full, it gains the client name, the measured outcomes and the lessons learned, following the template on the right.
Named references are available under NDA for qualified buyers; ask us during your evaluation.
Named references under NDA
Qualified buyers can speak to reference clients under NDA once the client has agreed. Allow a week or two for the introduction.
Other work we can describe
Beyond the four perspectives above, we have delivered cybersecurity remediation and SSH hardening across RHEL estates, Dynamics AX high availability with Hyper-V failover clustering, OT and ICS cybersecurity assessments for industrial sites in Saudi Arabia, SAMA and NCA ECC compliance work for Gulf financial clients, and an Arabic invoice extraction feasibility study using several vision-language models with a judge model. Our industries pages show where each kind of work tends to land.
Frequently asked questions
Why are there no client names or numbers on this page?
Because they are not ours to publish. Several of our clients are regulated financial and energy companies that do not permit public references, and others have simply not been asked yet. We would rather describe real work accurately without names than publish logos we cannot stand behind. Named references are available under NDA during an evaluation.
Can we speak to a reference client?
Yes, for qualified evaluations. Tell us the capability and industry that matter to you and we will approach the most relevant client for permission; most agree to a call under NDA. We will not promise a reference before the client has said yes, so allow a week or two for the introduction.
Are any of these engagements still running?
Some are in support or improvement phases, which is why we can describe the architecture with confidence: it is still in production. Where a system has been handed over fully to the client's team, the perspective reflects the state at handover. We update a perspective when the client approves further detail.
Do you have quantum case studies?
No, and we say so plainly. Our quantum work is research-led: feasibility studies, simulation and post-quantum cryptography readiness planning. There are no client quantum deployments to describe yet. The innovation page explains the current research threads and how a client can propose a challenge.
Discuss your challenge
If one of these engagements resembles your situation, tell us. We will connect you with the engineers who worked on it and, with permission, the client.