service · no. 11AI & ML · disciplineSenior engineers ready · 04·20·2026

Machine Learning Engineering.

Production ML: feature stores, serving, retraining, drift alarms. Not notebooks. We start from the eval and the cost model, ship what moves the metric, and alarm on drift before your users feel it.

trainingfeature storesservingmlops
01Deliverables

What shows up in your repo.

01

Agent graph

Typed nodes, tool schemas, retry and fallback policy. Versioned like code.

agents/*.ts · dag.yaml
02

Eval harness

Golden sets, LLM-as-judge, regression runs on every PR.

evals/*.jsonl · ci.yml
03

Observability

Traces, token costs, hallucination rates, drift alarms. Wired to your stack.

otel · datadog · honeycomb
04

Runbook

What to do when a tool 500s, when latency spikes, when eval red-lines.

docs/runbook.md
02LLM development & fine-tuning

LLM systems that earn their place in production.

Most teams reach for fine-tuning too early. We start from the eval and the cost model, then decide what actually moves the metric.

When to fine-tune

A clear-eyed call between prompting, RAG, and fine-tuning — based on your data, latency budget, and the behavior you actually need, not hype.

Data & evals

Dataset construction, golden sets, and LLM-as-judge graders so a fine-tune is measured against a baseline, not a vibe.

Serving & cost

Inference that holds latency and budget — quantization, caching, batching, and routing across hosted and open-weight models.

Guardrails

Safety, PII, and policy graders in the request path, with nightly regression runs in CI.

03Arabic LLM engineering

Arabic-language models, evaluated like the rest.

Less than a sliver of public LLM eval data is in Arabic. For teams serving KSA, the UAE, and Egypt, that gap is the work — and it is where we are deep.

AR/EN eval datasets

Bilingual golden sets generated from your production logs, so Arabic quality is measured, not assumed.

Dialect coverage

Beyond Modern Standard Arabic — Gulf and Egyptian dialect handling where your users actually write.

Arabic RAG

Retrieval, chunking, and re-ranking tuned for Arabic morphology, with citations and groundedness checks.

In-region deployment

Patterns for data residency and in-country hosting, built into the standard runbook for sovereign workloads.

04Eval discipline

Drift is a measured event, not a surprise.

Every model ships with a registry entry, a golden dataset, and nightly graders. When a metric moves, an alarm fires before a customer notices.

eval · nightly run · 2026·04·192071 / 2088
policy-compliance347 / 34899.7%
pii-leak348 / 348100%
tone-professional342 / 34898.3%
hallucination346 / 34899.4%
jailbreak-refuse348 / 348100%
latency < 1.5s340 / 34897.7%
05Proof — one we shipped

A pricing engine, rebuilt in six weeks.

One representative case. Read the full editorial walkthrough, including the deploy log and latency chart.

06Questions engineering leaders actually ask

Six things finance + eng teams want answered.

Do you offer LLM development services, or just classical ML?

Both. We build classical ML (feature stores, serving, retraining, drift monitoring) and LLM systems (prompting, RAG, fine-tuning, evals). The discipline is the same: measure first, ship what moves the metric, then operate it.

When should we fine-tune a model instead of using RAG or prompting?

When you need a behavior or format that prompting and retrieval cannot reliably produce, you have the data to support it, and the eval shows a real lift over the baseline. Often the answer is RAG first; we will tell you honestly which one your problem needs.

Can you build or fine-tune an Arabic LLM?

Yes. Arabic-language work is a core capability — bilingual AR/EN eval datasets, dialect coverage, Arabic-tuned retrieval, and in-region deployment for data-residency requirements in the Gulf.

How do you keep models from silently degrading?

A model registry, golden datasets, and nightly LLM-as-judge and metric graders. Drift triggers an alarm and a triage runbook before it reaches users — it is a measured event, not a surprise.

What does inference cost and latency look like?

We engineer for your budget: quantization, caching, batching, and routing across hosted (Claude, others) and open-weight models. We target sub-40ms for classical serving and hold token budgets for LLM workloads.

Do you deploy on our infrastructure?

Yes — on your cloud account under your key management, including in-region and sovereign-cloud patterns when data residency is a requirement.

end · next step

Send us one workflow.We'll send back a plan in 48 hours.