AI Systems Architect ● Open to Remote Roles

I build AI automation systems and CRM integrations that scale your business, not your headcount.

I'm an AI automation developer building multi-agent pipelines, full-stack CRM architecture, API integrations, voice AI agents, and custom dashboards, end to end, for businesses that are done operating manually. Four years of this across more than 30 companies, remote and worldwide.

0+ Years in AI & Automation
0+ Businesses Transformed
0+ CRM Platforms Mastered
0+ Open-Source Templates
HubSpot Salesforce GoHighLevel n8n Zapier Make OpenAI Anthropic Retell Vapi LangChain Zoho Pipedrive ElevenLabs Perplexity Mistral Airtable Notion ClickUp ActiveCampaign Python PostgreSQL Stripe Supabase React Gemini Pabbly CrewAI Monday.com
HubSpotSalesforceGoHighLeveln8nZapierMakeOpenAIAnthropicRetellVapiLangChainZohoPipedriveElevenLabsPerplexityMistralAirtableNotionClickUpActiveCampaignPythonPostgreSQLStripeSupabaseReactGeminiPabblyCrewAIMonday.com
Ammar Imtiaz, AI automation developer and CRM integration expert

Ammar Imtiaz

AI Agents & Agentic Workflows
API Integrations & ETL Pipelines
CRM Architecture & Migration
Workflow Automation Engineering
Voice AI Agents
Custom Web Applications
SEO Infrastructure & Tools
System Architecture & Design

Not a generalist. Not a consultant with a deck. A builder.

I've spent 4+ years at the intersection of AI, automation, and business operations, deploying real systems inside legal firms, medical practices, e-commerce brands, SaaS companies, and government contractors.

I work across every major CRM, every frontier AI model, and every automation platform worth knowing. The difference between me and most "automation experts" is that I don't just connect APIs. I architect the logic that makes entire businesses scale without adding headcount — no people in the loop.

"If you want someone who goes from whiteboard to production-ready system in days, not months. That's the work I do."

Systems that run your business. Not you.

Each one handles a full business function end-to-end — no human in the loop. Click any system to see exactly how it works and what it's built on, then see what else I build or what a build like this costs.

System / 05
BidStrike
A GovCon capture platform that finds federal contracts, scores your odds of winning, and drafts the compliant proposal — you just approve the submit.
What it does

BidStrike watches six government contract sources in one feed — federal, state, local, grants, and international — matched to your NAICS codes and set-asides (8(a), SDVOSB, WOSB, HUBZone). Every open solicitation that fits you shows up. No manual SAM.gov digging.

It doesn't wait for the RFP. Pre-RFP Radar surfaces agency acquisition forecasts and early signals 6–18 months before the solicitation exists — and drafts your capability response to sources-sought notices while competitors are still asleep on it.

Every contract gets a 0–100 win-probability score. PWIN plus an AI deal brief with a go / caution / no-go verdict, incumbent history pulled from public award data, and a recompete radar for expiring awards the agency has to re-buy. You stop burning weekends on bids you can't win.

Then it writes. Shipley-structured proposals — executive summary, technical approach, management plan, past performance, pricing — built from your company profile, with a compliance matrix pulled from the actual solicitation. Three AI review passes (Pink / Red / Gold) critique the draft before you ever see it. Export to Word and PDF.

On AutoPilot, it stages the submission to the contracting officer and waits for your approval. Nothing goes out without you. Clean, simple solicitations can auto-submit within limits you set; anything with attachments is held as a draft. Every action is permission-checked and audit-logged.

Free to start. No card. No demo call.

How it's built
01
Connect your profile
NAICS codes, certifications, past performance, and target agencies go in once. Opportunity filters and alert preferences set your feed.
02
AutoPilot finds it
Monitors SAM.gov and five other sources 24/7, scans thousands of daily solicitations, and fires instant alerts with pre-scored matches.
03
Score and qualify
PWIN plus compliance scoring on every contract. Incumbent and competitive-field data from public awards. Go / no-go decided on data, not a gut call.
04
Build the proposal
AI-assisted Shipley drafts, auto-generated compliance matrix, requirement extraction, and three color-team review passes before human eyes.
05
Track to award
Pipeline dashboard for submissions, follow-ups, and awards. Win/loss data feeds back into the next capture. Post-award: CPARS prep, option-exercise deadlines, recompete radar.
Stack
Next.js Claude API Supabase Clerk Stripe Resend SAM.gov Grants.gov SBA DSBS Word/PDF Export
Start free →
System / 04
Outreach System
A 7-day follow-up machine that calls, texts, and emails every lead — so you never miss a close because someone forgot to follow up.
What it does

The moment a lead comes in, a 7-day sequence kicks off automatically across call, SMS, and email. You give us a valid phone number and email. We handle everything after that.

Days 1–3 are aggressive. A call goes out immediately, followed by an email and SMS. If there's no response after 5 hours, it tries again — call and SMS. Next day: another round. Day after: one more.

Days 4–7 shift tone. One email and one SMS per day, written to sound like it's coming from you personally — not a bot. A call is only triggered if they open an email and don't reply within an hour.

At any point: if a lead replies, you get notified instantly. If they say STOP, they're removed immediately. If they book an appointment, you're notified. If they open an email but ghost it for an hour, the system calls them automatically.

Three sequence styles available — same price, different levels of aggression — built for different markets and sales cycles.

How it's built
01
Trigger & Contact Intake
Lead enters via form, CRM, or webhook. GoHighLevel captures the contact and fires the automation workflow instantly — no delay, no queue.
02
Sprint Phase (Days 1–3)
GHL workflows handle call queuing via VAPI or Retell AI for outbound voice. Email and SMS send through GHL's built-in channels. 5-hour wait timers are set inside the workflow. Each branch checks for a reply event before firing the next step.
03
Nurture Phase (Days 4–7)
Personalized messages are generated using Claude — given the lead's name, business, and context from the CRM. Email open tracking is handled by GHL's pixel. If open detected and no reply for 60 minutes, the workflow triggers an AI call via VAPI or Retell.
04
Reply & Intent Detection
All inbound replies feed into Claude via the GHL webhook. It classifies intent: hot (wants to talk), neutral (question), STOP (unsubscribe). Routes accordingly — notifies owner, books calendar, or removes contact.
05
Notifications
Every significant event — reply, booking, unsubscribe — fires an instant email notification to the business owner via GHL's email action node. No dashboard to check. You get an email.
Stack
GoHighLevel VAPI Retell AI Claude API n8n
System / 03
Smart Reputation Management
Good reviews get through fast. Bad ones get caught, handled personally, and never reach the internet.
What it does

After a job or interaction, the system reaches out to the customer by email or SMS asking how their experience was. The customer replies in plain language — the AI reads what they said and decides what happens next.

If they're happy, the AI sends them a direct link to leave a public review — removing all friction between a satisfied customer and a Google review.

If they're unhappy, no review link ever appears. Instead, they get a personal apology and the business owner is notified so they can step in, resolve it, and potentially turn the situation around before it becomes a public problem.

The system also tracks whether the customer even opened the first outreach email. If they didn't open it, it tries again — up to 3 more times. If they did open it but didn't respond, it follows up 4 times — because someone who opens but doesn't respond is worth nudging.

How it's built
01
Post-Interaction Trigger
Triggered by a GHL pipeline stage change, a completed appointment, or a manual tag. Sends an outreach email or SMS through GHL asking for feedback.
02
Email Open Tracking Logic
GHL's email open pixel fires when the email is viewed. The workflow checks open status after a set delay — if not opened, re-sends up to 3 times. If opened with no reply, triggers up to 4 follow-ups.
03
Sentiment Classification
Customer reply hits a webhook. Claude receives the raw reply with a classification prompt — outputs: positive, neutral, or negative intent. No keyword matching. It reads what was actually said.
04
Routing
Positive → GHL sends the Google review link automatically. Negative → Claude drafts a personalized apology. GHL sends it and simultaneously fires a notification to the business owner with the customer's exact response.
Stack
GoHighLevel Claude API n8n
System / 02
Content Engine
From market research to published blog, LinkedIn post, tweets, images, and video ads — a full content pipeline that runs on its own.
What it does

The engine starts by figuring out what your market actually cares about. It finds the 5 biggest pain points in your industry and as many working marketing angles as it can pull — then scans competitors to understand how they write, what they say, and how their ads look.

From that research, it produces five written reports: your brand voice, your positioning, where competitors sit, the copy style that works in your market, and how you should differentiate.

It then goes into planning — taking each pain point and marketing angle and generating 5 content ideas around each one. It picks the best 3 and finds proven hooks for each of them.

From there, a content writer produces a 250-word SEO blog post for each idea, a LinkedIn post referencing it, and a set of tweets built around the same topic — written as independent observations, not promotions.

Then the media side kicks in: 5 static ad images, 1 UGC-style video, and 5 video ad scripts. You pick one script. The video gets produced using AI video generation — delivered and ready to post.

Throughout all of this, Claude and Grok are reading each other's output and debating what's good enough before anything moves forward. Perplexity is the search engine for everything. If something can't be found there, the SERP API is the backup.

How it's built
01
Research Agent
Claude and Grok both query Perplexity API for industry trends, pain points, and competitor positioning. Where Perplexity hits its limits, SerpAPI fills the gap with live search results. Both models debate and consolidate findings before the next stage starts.
02
Reporting
Claude takes the consolidated research and writes five structured reports — brand voice, positioning, competitor map, copy style, differentiation angle. These documents feed every downstream step and are stored for reuse.
03
Content Planner
A planning agent loops through each pain point and marketing angle, generating 5 content ideas per input. Claude and Grok independently score each idea against engagement criteria. The 3 highest-scoring ideas per angle advance.
04
Hook Research
For each approved idea, Perplexity searches for high-performing hooks from top content in the same niche. Claude selects and adapts the best fit to the brand voice.
05
Content Writing
Claude writes the 250-word SEO blog, LinkedIn post, and standalone tweets — separate prompts, each with different tone and structure constraints. The blog targets keyword density and readability.
06
Media Generation
Static images: Nanobanan generates base visuals with ad-style overlays layered programmatically. UGC: Nanobanan paired with HeyGen API for avatar-based video. Video ads: Veo 3 and Kling 3.0 via kie.ai API generate the final video ad from the approved script.
Stack
Claude API Grok API Perplexity API SerpAPI Nanobanan HeyGen API Veo 3 Kling 3.0 kie.ai API n8n Make
System / 01
Chat & Call Agents
Inbound and outbound agents that talk to people, remember what was said, and handle bookings — across every channel at the same time.
What it does

One agent handles every conversation channel at once — email, SMS, WhatsApp, Facebook, Instagram, live chat on your site — and an AI voice agent handles calls. They all know who they're talking to and what was said before, regardless of which channel the person switches to.

For inbound: someone messages on Instagram and asks about availability — the agent answers, checks the calendar, and books the appointment without you touching it.

For outbound: the system follows up with people who didn't respond, re-engages old leads, and reminds people about upcoming appointments.

At any point during a chat or call, the agent can reschedule, cancel, or book a new appointment directly — no hold music, no "I'll pass this along." It does it live. If the conversation reaches a point where a human needs to step in, the agent flags it and hands off cleanly.

Issues get solved. Questions get answered. Appointments get managed. None of it requires you to be online.

How it's built
01
Unified Conversation Layer
GoHighLevel's conversation inbox aggregates all channels — SMS, email, Facebook Messenger, Instagram DM, WhatsApp, and live chat — into a single thread per contact. Every message in or out goes through one API surface, so the AI always has full context regardless of channel.
02
AI Chat Agent
Inbound messages trigger a GHL webhook to n8n. n8n passes the full conversation history and contact data to Claude. Claude generates a response based on a business-specific system prompt — FAQ, services, pricing, tone. Response posts back through GHL's messaging API to the original channel.
03
Calendar Integration
Claude has access to GHL's Calendar API as a tool call. When a user asks to book, reschedule, or cancel, the agent checks real-time availability and completes the action mid-conversation. Confirmation goes back to the user in the same message thread.
04
Voice Agent
VAPI or Retell AI handles inbound and outbound calls. The voice agent shares the same knowledge base and CRM data as the chat agent — it knows who it's calling, the history, and the intent. Post-call transcript syncs back to the GHL contact record.
05
Follow-up & Handoff
If a conversation goes cold, n8n fires a re-engagement workflow after a set interval. If Claude detects escalation signals — anger, legal language, complex issues — it tags the conversation, pauses the AI, and sends the business owner an alert with full context to take over.
Stack
GoHighLevel Claude API VAPI Retell AI n8n

My everyday building blocks.

Browse the n8n library →

What I can build.

Six things, built end to end rather than advised on. If you want the proof before the list, the systems already running in production are there, and the cost, timeline and ownership questions are answered further down.

Custom Web Apps & Dashboards

Full-stack web applications with real-time data, clean UI, and direct API access to your stack. Dashboards that replace 10 browser tabs with one command center.

React · Python · Supabase · PostgreSQL · REST APIs · Stripe

Multi-Agent AI & Agentic Workflows

I design and deploy multi-agent systems using LangChain, CrewAI, and custom frameworks. agents that research, decide, and act autonomously with defined guardrails.

LangChain · CrewAI · AutoGen · n8n AI Agents · OpenAI · Claude API

CRM Architecture & Integration

Full CRM setup, migration, and multi-platform integration. I've worked inside 10+ CRMs and can wire them together, migrate data without loss, and build automation on any stack.

GoHighLevel · HubSpot · Salesforce · Zoho · Pipedrive · Close · ActiveCampaign

API Integrations & ETL Pipelines

Anything with an API can be connected. I build fault-tolerant data pipelines that move, transform, and sync data across your entire stack. zero manual entry.

REST APIs · Webhooks · Python · Supabase · PostgreSQL · Airtable

Advanced Workflow Automation

Complex, multi-step workflows that cover every edge case. I migrate fragile Zapier/Make flows into robust, self-healing n8n infrastructure built to scale.

n8n · Make · Zapier · Pabbly · Custom Triggers · Webhooks

Voice AI Agents

Hyper-realistic AI voice agents for inbound support and outbound sales. booking into your calendar, qualifying leads, and handling objections at scale.

Retell · Vapi · Assistable · ElevenLabs

What operators say after we've shipped.

What people ask before they hire me.

Scope, cost, timelines, and the questions worth asking any automation builder before you sign. Answers come from systems already running in production, not from a services page. If yours is not here, ask it directly.

What does an automation project cost to build?

Ranges are wide because "automation" covers everything from a single Zapier fix to a system that runs a whole business function. A one-off workflow lands in the hundreds. A real build, meaning integrations, error handling, and something your team can operate without me, typically runs 3k to 10k depending on how many systems it touches and how bad the data is going in. I quote after seeing your stack, never before, because most of the cost sits in edge cases nobody raises on a first call.

How do you price: fixed fee, hourly, or retainer?

Fixed fee for anything with a definable end, which is most builds. You get a number before work starts, and if I estimated badly that is my problem rather than your invoice. Hourly only for exploratory work nobody can honestly scope yet, usually an audit of an existing mess. Retainers make sense in exactly one case: the system genuinely changes month to month and you want someone holding it. If your system is stable and someone is selling you a retainer, they are selling you insurance you probably do not need. Scope changes get repriced in the open, not absorbed quietly and resented later.

Should I hire an AI automation agency or an independent developer?

It depends on whether your problem is technical or organisational. An agency earns its fee when several disciplines have to run in parallel, design and copy and paid media alongside the build, or when procurement needs a company with insurance and a support agreement. The tradeoff is that the person who sold you the project is usually not the person building it, and the build lands with whoever happens to be free that week. An independent builder is the better bet when the work is genuinely technical, because you talk to the person writing the workflows and nothing is lost in the handoff. Ask either one the same question: who specifically writes this, and can I speak with them before I sign.

What does an AI automation consultant actually do?

The useful version audits where your operation loses hours to manual coordination, decides which of those are worth automating and which honestly are not, then builds and hands over the systems. That last part separates consultants from contractors. Plenty will produce a recommendations deck and leave you with the original problem plus an invoice. Ask whoever you are evaluating to show you a system they built that is still running a year later, and to tell you who maintains it now. If that answer is vague, you are buying slides.

What should I look for in an AI agent development company?

Production evidence, not demos. Ask what happens when the model returns something unexpected, where tool calls get logged, how spend per run is capped, and exactly which actions need a human to approve. Each of those has a specific answer if the team has shipped agents against real systems, and a hand-wave if they have only built prototypes. Ask what they refuse to automate, too. Anyone who says yes to everything has not had an agent running against a live CRM at 2am. Framework choice, LangChain or CrewAI or something custom, matters far less than whether the guardrails exist at all.

What is custom AI agent development?

Building an agent that has tools, memory, and permission boundaries, rather than a chatbot answering from a prompt. The agent gets a goal, a defined set of actions it may take against your systems, and constraints on what it can do without approval. In practice that means LangChain, CrewAI, or a custom framework, plus the unglamorous parts: logging every tool call, capping spend per run, and deciding exactly where a human signs off. Off-the-shelf agents demo beautifully. The difference in production is entirely in the guardrails.

How do you migrate a CRM without losing data?

You never migrate straight across. Export, map fields into a staging layer, then run a dry pass against a sandbox so you can see exactly what breaks before anything is live. Deduplication rules get agreed before records move, because two systems always disagree about the same contact. Custom fields, activity history, and file attachments are where migrations quietly lose data, so those get validated by row count and by spot check. Both systems run in parallel until the numbers reconcile. Cutover is the boring part when the mapping was done properly.

Which CRMs do you integrate?

GoHighLevel, HubSpot, Salesforce, Zoho, Pipedrive, ActiveCampaign, Close, Microsoft Dynamics, and Monday or Airtable when a team is using one as their CRM. Integration work is rarely limited by which CRM you picked. It is limited by what the API exposes and how aggressively it rate limits. Salesforce and Dynamics expose almost everything and charge for the privilege. GoHighLevel and Pipedrive are faster to wire but have gaps you design around. If a platform has an API it can be connected. The real question is what keeping that connection stable costs.

What does a Salesforce integration project actually involve?

Far less API work than people expect and far more decision work. Salesforce exposes nearly everything through its REST, Bulk, and Streaming APIs, so connecting it is rarely the hard part. The hard part is agreeing which system owns a record, deciding what happens when both sides edit the same field, and handling the objects your org has customised over a decade. Sandbox first, always, because validation rules and required fields will reject writes that passed cleanly in your test payload. Watch the daily API call allocation on lower editions too. A sync built without batching can burn through it before lunch.

How long does a CRM integration project take?

One clean integration between two systems with decent APIs is usually days, not weeks. Full CRM architecture, meaning migration plus multi-platform sync plus the automation layer on top, runs two to six weeks depending on data quality and how many people have to agree on field definitions. The build is rarely the slow part. Waiting on API credentials, settling which system is the source of truth, and cleaning historical records account for most of the calendar. Projects that slip almost always slipped on decisions, not on code.

What do API integration services actually cover?

Connecting systems that were never designed to talk, then keeping that connection alive. Concretely: authentication and token refresh, mapping fields between two different data models, handling rate limits and pagination, retry logic with backoff, a queue so nothing is lost while the other end is down, and alerting that names the record which failed instead of just announcing that something did. The connection itself is often a day of work. Everything listed after it is what separates an integration that survives a year from one that stops silently on a Tuesday. If a quote never mentions error handling, you are buying the first day only.

Zapier, Make, or n8n: which should a growing business use?

Zapier when the flow is short, linear, and needs to be live today. Make when the visual layout genuinely helps whoever maintains it and the branching is moderate. n8n once you have real conditions, need self-hosting for data residency, or your Zapier invoice has started to resemble a salary. Migrating Zapier into n8n is the single most common request I get, usually triggered by cost or by one fragile flow failing silently for a month. Self-hosted n8n has no per-task pricing, which is the whole argument at volume.

What does self-hosted n8n actually cost to run?

The server, and very little else. A small VPS handling a few thousand executions a day sits somewhere between 10 and 40 dollars a month, and self-hosted n8n charges nothing per task, which is the entire argument once volume climbs. n8n Cloud is the better choice below a few thousand executions a month, or when nobody on your team wants to own a server. The cost people forget is upkeep: version upgrades, moving to queue mode when one instance stops coping, backing up the workflow database, and someone noticing before the disk fills. Budget for that rather than for a licence. A Zapier bill in the high hundreds usually pays for a self-hosted setup within a quarter.

Do I need a GoHighLevel developer, or can my team configure it?

Most of GoHighLevel is configuration and your team should own it. Pipelines, calendars, forms, standard workflows, and email sequences are all point and click, and paying someone to click on your behalf is waste. You need a developer when you push past what the builder exposes: custom values driving logic across sub-accounts, the v2 API, webhooks into outside systems, snapshot deployment across many locations, or a data migration in either direction. The other case is an account that has grown into hundreds of workflows nobody can map any more. That is an audit problem rather than a build problem, and it is worth paying to untangle before it breaks.

Can an AI voice agent really book appointments and handle objections?

Yes, within a narrow scope. Voice agents reliably qualify callers, answer questions from a known set, check a live calendar, and book, reschedule or cancel inside the same conversation. They handle common objections when those responses were scripted from real call recordings. What they do not do well is open-ended negotiation or anything needing judgment about an unusual account, so the right design routes those to a human mid-call instead of pretending. Latency matters more than voice quality. A pause beyond roughly a second reads as a bad line.

Is this RPA, or something different?

Different, and the distinction changes what it costs you. Classic RPA drives the user interface, clicking through screens the way a person would, which is exactly why it breaks the week a vendor moves a button. What I build talks to APIs directly, so it survives interface changes and runs in milliseconds rather than seconds. RPA still earns its place against genuinely legacy systems with no API at all, mainframes and old desktop software, and there I would reach for it rather than pretend otherwise. If your systems expose APIs and someone is quoting you screen automation, ask them why.

What happens when an API changes and the automation breaks?

It will break eventually, so the system gets built expecting it. Every integration has retries with backoff, a dead letter path so failed records are held rather than dropped, and alerting that names which step failed and with what payload. Silent failure is the actual risk. A flow that quietly stops routing leads costs far more than one that crashes loudly on day one. Handover includes what to check first when something breaks, so your team is never dependent on me being awake to diagnose it.

Do I own the systems you build?

Yes. Workflows, code, prompts, and documentation are yours, running on your infrastructure under your accounts. Self-hosted n8n sits on your server. API keys are issued from your accounts, never mine. That matters more than it sounds, because a lot of automation work gets delivered inside an agency's own tooling, which means the day the relationship ends, the system ends with it. Ask anyone you are evaluating where the workflows physically run and whose name is on the API keys.

If you would rather see the work than read about it, browse what I build or book a call.

Let's build something serious.

I take on a limited number of engagements each month. Whether you need a consultant, a contractor, or a full system architect. Let's talk.