Ahmad Tayel · Data and AI

Data and AI systems built to compound, not decorate.

10+ years of expert practice across data, AI, and engineering. Built for fintech, healthtech, and growth teams that ship. Currently based in Spain, available for fractional engagements through Consultalyst.

Book a 25-Minute Intro See Selected Work

Services

At the intersection of Data, AI, and Consulting.

Three ways an engagement lands. Each one starts with a written success contract before any work begins, and ends when the team can operate what was built without us.

Data Infrastructure and Analytics

Build the warehouse, pipelines, and dashboards your team actually uses. From event taxonomy to executive views.

Data warehouse architected on BigQuery, Snowflake, or PostgreSQL
Pipelines in dbt + Airflow with documented modeling layer
Tableau or Looker dashboards your execs open weekly

How It Ran

A Series B fintech needed a single source of truth across 5 product surfaces. Engagement: 8 weeks, ended with a documented dbt project, modeled warehouse, and 3 executive dashboards the team owns.

AI Workflows and Production Agents

From idea to a production AI workflow with the eval suite that decides what actually ships. Eval-first, not demo-first.

Working AI workflow integrated into your existing stack
Eval suite covering accuracy, latency, cost, and known failure modes
Runbook and monitoring so your team operates it without us

How It Ran

An SMB ops team wanted to automate ticket triage. Engagement: 6 weeks, ended with a deployed LLM workflow + eval bench catching 92% of mislabel cases before deploy.

Product Analytics and Experimentation

Instrument the funnel, design the A/B test, read out the result. Stop shipping on vibes. Decisions backed by evidence.

Event taxonomy and instrumentation across product surfaces
A/B test design + read-out framework your PMs can run independently
Cohort, retention, and funnel models surfacing signal, not noise

How It Ran

A growth team's activation rate was flat for 6 months. Engagement: 10 weeks, ended with 12 funnel events instrumented, 3 A/B tests run, and a documented 18% activation lift.

Book a 25-Minute Intro

Selected Work

Engagements where instrumentation became leverage.

Three pieces of work, anonymized by client. Numbers shown are what the teams measured themselves. Named attribution on the resume.

Cross-industry, fintech and SaaS

Production AI Eval Systems for LLM Workflows

Eval suites that catch model regressions before deploy. Covers accuracy, latency, cost, and known failure modes. Teams ship faster because the test bench is honest.

Problem: Teams shipped LLM features on vibes. No way to tell if a prompt tweak made things better or worse, and no agreed bar for 'good enough'.
Approach: Designed eval datasets from real production traces, set thresholds with the product owner, automated the run on every deploy with explicit pass / fail gates.

Python
Anthropic API
Pydantic
GitHub Actions

MENA buy-now-pay-later platform

Funnel Instrumentation and A/B Test Design

18% fewer drop-offs at the most-leaked funnel stage. Around 200 additional activated accounts per month, attributable in the warehouse.

Problem: A 40%+ drop-off between funnel start and activation with zero visibility into where or why people churned out.
Approach: Instrumented 12 stage events with consistent taxonomy, ran cohort analysis to surface the three biggest friction points, designed A/B tests for each.

Amplitude
SQL
Tableau
A/B Testing

Payment processing, multi-tenant

Real-Time Operations Dashboards That Cut Detection Time

Gateway failure detection moved from hours to minutes. Flagged 200+ suspicious merchant accounts through transaction pattern analysis the ops team could act on directly.

Problem: Outages were noticed manually, often hours after they started. Ad-hoc reports took days to turn around for any operational question.
Approach: Built a real-time success-rate dashboard with parameterized alerting thresholds. Stored procedures replaced the ad-hoc SQL grind so the team could self-serve.

Tableau
SQL
Stored Procedures

Expert practitioner. Built for teams that ship.

10+ years of expert practice spanning data engineering, product analytics, machine learning, and AI workflow builds. Sectors covered: fintech, healthtech, payments, subscription, and growth-stage SaaS, primarily across MENA and Europe.

Currently completing an MSc in Data Science and Artificial Intelligence at the University of Liverpool. Consultalyst is the brand for fractional engagements. Sister practice Zajel AI ships production AI workflows for SMB teams.

10+

Years of expert practice across data, AI, and engineering.

5M+

Users served via production analytics pipelines.

MSc

Data Science and AI, University of Liverpool.

Also Runs

Zajel AI for production AI workflow builds.

When a brief needs production AI shipped, not just designed, Zajel AI carries it. Sister practice, dedicated build team, same standards.

Visit Zajel AI

Intelligent WorkflowsAutomation across ops, support, and internal tooling.
Data-Driven ProductAI features grounded in product analytics, not vibes.
SMB ScaleEnterprise-grade engineering without the overhead.

Work With Me

Start with the problem.

Three fields, one 25-minute intro call if it's a fit. Every submission gets a personal read.

Prefer email? hello@consultalyst.com