haithams.info
LinkedIn GitHub Email
Haitham Yasser Al MaamariMuscat, Oman
Haitham Al Maamari speaking at an Omantel event

Data analytics / AI platforms / automation

I translate business requirements into SQL reporting, data-platform migrations, and practical LLM and automation solutions.

Explore selected work
Current roleAssistant SpecialistData Analytics & AI Platforms, Omantel
Current focusApplied LLM workflowsPractical automation with clear validation
Portfolio contentsLive and internal workSystems, methods, and professional profile

01 / Selected work

Proof over promises.

Every project starts with a practical problem. The work below shows how I approach data and AI delivery: clear inputs, explicit ownership, deterministic checks, and outputs people can use.

01Automation system

Live / private deployment

A live daily briefing with validation built into delivery.

A VPS-hosted workflow runs at 9:00 AM, uses Hermes to gather candidate stories, then passes structured JSON to a Python layer that validates, ranks, archives, and delivers a briefing to approximately 10 recurring users.

Phase 01

Gather

Hermes gathers candidate stories from a source catalogue and writes structured JSON for the deterministic pipeline.

Five mandatory checks

V1 - V5
  1. Freshness and sourceRequires valid source metadata and a publication date inside the freshness window.
  2. Cross-referenceHigh-priority stories require an independent confirming source.
  3. Archive deduplicationChecks the previous three issues before repeating a story.
  4. Internal consistencyRejects duplicate headlines, duplicate URLs, and invalid URLs before ranking.
  5. Editorial countsRecords candidate, pass, and drop counts for the operator.

The public package exposes the sanitised validation, ranking, formatting, archive, and delivery layer while private prompts, recipients, credentials, and VPS configuration remain excluded.

View public repository
  • Python
  • Hermes
  • VPS
  • Deterministic validation
02AI assistant

Archived prototype / ranked 1st

Context before the reply.

An LLM-powered assistant built to read an entire email thread, extract relevant context, and draft a response with configurable controls. The prototype was archived after Omantel adopted Microsoft Copilot.

1st

Ranked 1st in Omantel's Gen Z AI project competition.

02 / Enterprise work

Systems work, with the data kept private.

My Omantel work involves internal data and cannot be publicly demonstrated. My responsibilities include translating stakeholder requirements into SQL reports, rebuilding dashboards against the new data warehouse, migrating scheduled reports, preparing governance datasets, and coordinating cross-team delivery.

Live / internal collaboration

Telecom competitor intelligence portal

Collected and validated competitor data, gathered stakeholder requirements, and contributed to interface development. The collaborating team remains responsible for automation, deployment, and ongoing support.

SQL reporting for campaigns, promotions, operational analysis, and ad hoc requests Dashboard rebuilding during migration from the legacy reporting platform Scheduled-report migration and platform-transition continuity Data governance, ORKI initiatives, and stakeholder coordination

03 / Practice

A broad toolkit, held together by method.

01

Data reporting

SQL, Greenplum, DBeaver, dashboard rebuilding, scheduled-report migration, and requirements gathering.

02

Applied AI

Generative AI, prompt engineering, LLM workflow and assistant design, Anthropic Claude API, local Qwen3-8B, and Hermes.

03

Automation

n8n, cron scheduling, multi-source collection, content validation, archive deduplication, and messaging automation.

04

Prototyping and ML

LangChain, LangGraph, model routing, Python, supervised and unsupervised learning, NLP, and time-series forecasting.

04 / Profile

A career built through changing contexts.

  1. 2025 - now

    Assistant Specialist

    Omantel, Data Analytics & AI Platforms

  2. 2024 - 2025

    Gen Z Trainee

    Omantel, Data Science & Machine Learning Programme with Codeline

  3. 2022 - 2024

    Intern

    Jarzez Trading LLC, operations and project coordination

  4. 2017 - 2022

    BSc Computer Science

    University of Reading, United Kingdom

Languages Arabic (Native) / English (Fluent)