Skip to content
Available for projects / rolesAI Automation · Chatbots · Full-Stack Engineering

Leads answered in seconds, not hours.

I build the agents and the pipelines that run them. WhatsApp inbound → intent detection → guardrails → quoting → CRM sync → human handoff. Deployed inside your existing stack, measured at every step.

See the pipelineStart a conversation
  • 8smedian lead response
  • 60%resolved with no human in loop
  • 5agents running in production

pipeline · inbound message path

0 events

POST /webhooks/whatsapp/messages

sig verify · HMAC-SHA256 ok

enqueued → dq:inbound · 210ms

310ms p50 · 0 dropped

Selected Work

Systems that ship, end-to-end

Each project went from prompt architecture to database schema — deployed to real traffic with retry policies, guardrails, and an observability path before it mattered.

In production
01 / 04

WhatsApp AI Sales & Negotiation Agent

An always-on AI seller that qualifies, negotiates, and closes deals on WhatsApp, Instagram, and web chat.

ProblemSales teams were losing deals to slow manual follow-up — leads waited hours for quotes, price negotiations were inconsistent across reps, and no conversation ever made it into the CRM.

Meta Cloud APIPrompt EngineeringWebhooksFunction CallingCRM SyncNode.js

Architecture

01Meta Cloud API webhook receives inbound messages from WhatsApp

02Normalization layer → unified omnichannel inbox (WhatsApp, IG, web chat)

03Prompt engine with persona, discount guardrails & escalation rules

04Function calling → quote builder, CRM upsert, calendar booking

05Human handoff triggered on high-intent or risk signals

06CRM sync to HubSpot / Sheets with full conversation memory

Live DemoSource
In production
02 / 04

Automation & Orchestration Pipeline

Event-driven workflows that move data across business tools on their own — with retries, alerts, and full audit trails.

Architecture

01Triggers: webhooks, cron schedules, DB change events → entry gate

02n8n workflows containerized with Docker Compose

03Typed payload contracts, transformation & routing layer

n8nDockerWebhooksREST APIsPostgreSQLObservability
Live DemoSource
In production
03 / 04

Custom Web Platforms & Dashboards

Role-aware internal tools and realtime dashboards that turn scattered data into one trusted source of truth.

Architecture

01Next.js 15 App Router + React Server Components

02Typed data layer over REST endpoints with caching

03Supabase / PostgreSQL with row-level security & RBAC

Next.jsReactTypeScriptTailwind CSSPostgreSQLshadcn/ui
Live DemoSource
In production
04 / 04

RAG Knowledge Base & LLM Evaluation Lab

Multi-document retrieval-augmented chat with a test-first eval harness that catches regressions before users do.

Architecture

01Ingestion → chunking + embeddings → pgvector store

02Hybrid retrieval (vector + keyword) with reranking

03Grounded generation with citations & no-answer fallbacks

RAGpgvectorLLM EvaluationGuardrailsPrompt EngineeringCI/CD
Live DemoSource

Capabilities

One system, three disciplines

An AI layer that decides, a backend layer that executes, a frontend layer people actually use. Each one measured against a regression gate before it ships.

[ module-01 ]

AI & Agents

Prompt-to-production discipline: guardrails, evals, and function calling on every deployment.

  • System Prompt Engineering
  • Guardrails & Safety
  • Function Calling
  • RAG & Vector Search
  • LLM Evaluation
  • Agent Orchestration

status: 6 skills · in production · eval-gated

[ module-02 ]

Backend & Automation

Event-driven integration with typed contracts, idempotency, and a dead-letter path on everything.

  • n8n
  • Webhooks
  • REST APIs
  • PostgreSQL
  • Supabase
  • Docker
[ module-03 ]

Frontend & UI

Type-safe React interfaces held to a measurable performance and accessibility budget.

  • Next.js
  • React
  • TypeScript
  • Tailwind CSS
  • shadcn/ui
  • Framer Motion

$stack --summary --production

  • domains.count() = 3
  • skills.count() = 18
  • in_production = true

exit 0

About

Engineering with intent

I build agents and the systems that run them. The work sits at the seam between AI and engineering: autonomous agents that qualify leads and negotiate deals inside WhatsApp, pipelines that move data between tools with no human in the loop, and full-stack platforms that turn scattered operations into one source of truth. Every layer gets the same discipline — a prompt is reviewed like a schema, a webhook is designed like an API contract, and nothing ships without an eval, a retry policy, and a way to watch it fail.

The numbers matter more than the stack: median lead response cut from 45 minutes to seconds, roughly 60% of inbound answered with zero human touch, operational hours returned to teams every week. I prototype fast, then industrialize deliberately — typed contracts, idempotency, dead-letter handling, guardrails, and dashboards. Measure it, then make it faster.

01Measure the business impact, not the demo
02Every webhook is designed like an API contract
03Design for retry and idempotency from line one
04Nothing ships without an eval gate and guardrails

Timeline

  • 2023 — Present

    AI Automation Lead / Full-Stack Engineer

    Voltiflow Ltd

  • 2021 — 2023

    Founding Engineer

    Freelance Studio

  • 2019 — 2021

    Full-Stack Developer

    Momentum Digital

based: Remote · Worldwide

Contact

Tell me the metric you want moved

An agent, a pipeline, or a platform — describe the operation and the number you want to change. I reply within 24 hours.

Emailadam.aboali.aaa@gmail.com
GitHubLinkedInX / TwitterEmail

Contract, full-time, or a founding-partner seat — open to all of it. Lead with the problem and the current numbers; the stack is the easy part.