Fakhar E Ali FakharEAli

Fakhar E Ali

AI Systems Architect · Founder, Workup Solutions

What I build: multi-tenant SaaS, AI agents and automation fleets that run real businesses, in TypeScript, Next.js, Supabase/Postgres and n8n. How: agentic engineering with Claude Code, spec first, verification before ship, deterministic gates behind every model. Proof: 14 systems documented with real screenshots in the portfolio hub. About 95% of my commits live in private production repos; the case studies are the tour.

🏛 What I build

Pillar	What it means	Flagships
📞 AI Voice SaaS	Multi-tenant voice agents that answer a business's phone, book the appointment and enforce industry compliance deterministically	VerticalVoice
🎓 Education-vertical AI agents	Grounded assistants for training providers and academies: funding checks, level placement, enrolment, nurture sequences, with deterministic engines behind every model	SeatWise · CohortPilot
🎬 Content-agency operating stack	The analytics, clip-selection, creative-QC, prospecting and roster-management systems a content agency actually runs on	AlphaOS · Clipper Tracking · Carousel Lab · Hook + Caption Lab

🖼 Featured systems

VerticalVoice 🟢 Live _{Multi-tenant AI receptionist for healthcare, restaurants and real estate: one compiler, three industry packs, compliance enforced before any tool runs.}	AlphaOS 🟢 In production _{Operating system for a premium content agency: cross-platform analytics, 25 AI intelligence modules, a benchmarked clip extractor, meeting knowledge and pre-sales tooling.}	SeatWise 🟡 Pilot _{A sourced, dated, rule-cited net fee for funded courses in seconds, and the seat booked in the same conversation.}
CohortPilot 🟡 Pilot _{The admissions officer that never sleeps: grounded answers, weighted level placement, kanban pipeline and conversation-driven nurture sequences.}	Clipper Tracking 🟢 In production _{Mobile-first PWA running a freelance clipper roster: post calendar, labelling, trial-reel routing, weekly PDF report, champions and streaks.}	Carousel Lab 🟢 In production _{Transcript to on-brand Instagram carousel in about two minutes: AI decides, a deterministic renderer draws, nine QC gates check, a human approves.}

🧰 More systems

System	What it does	Stack	Status
Automation Fleet	36 n8n workflows across two self-hosted instances: onboarding, meeting intelligence, exactly-once reminders, prospecting, publishing, syncs	n8n · Supabase · Google Workspace · Docker · Traefik	🟢 In production
Hook + Caption Lab	Generator and critic agents for on-screen hooks and client-voice captions; a written standard compiled into deterministic gates plus a blind eval harness	Next.js · pgvector · Gemini · PM2 · Traefik	🟢 In production
Lead Engine	Two-arm lead engine: podcast clip-gap discovery plus real-time intent monitoring, ICP scoring with tiers, AI openers every morning	Python · Next.js · Supabase · Apify · Claude · Gemini	🟢 In production
Outreach Dashboard	Dark-first KPI dashboard for a three-person outreach team: 8-stage funnel, trends, leaderboard, goal pacing, CSV and PDF export, RLS-enforced ownership	Next.js · Supabase · Recharts · Framer Motion	🟢 In production
Social Connect	Signed-link creator onboarding: the agency mints an HMAC-signed link, the creator connects Instagram, Facebook, YouTube and TikTok in four clicks	Next.js · HMAC · Docker · Traefik	🟢 In production
Podcast to Clips	One episode in, 10 to 12 finished YouTube videos out: transcribe, segment, cut, thumbnail, copy, review, publish, learn	Node/TS · Express · Postgres · BullMQ · Deepgram · Gemini · ffmpeg	⚪ Internal
Viral Script Generator	Ten ranked short-form scripts in a creator's own voice: ingest reels, extract patterns, build a voice profile, generate with a critic loop	Next.js · pgvector · Claude · Gemini · Cohere · Deepgram	⚪ Internal
Autopost AI	v1 social scheduling and analytics with a clipper leaderboard and six-role access; the AlphaOS predecessor	Express · Prisma · Supabase · S3 · Gemini · PM2	⚪ Superseded

→ All 14 case studies

📈 By the numbers


14 systems documented, every one shipped to production or pilot	60 real screenshots (seeded demo data) across the case studies
3 live products you can open today: VerticalVoice, SeatWise, CohortPilot	36 n8n workflows running around the clock across two self-hosted instances
25 AI intelligence modules behind a versioned prompt registry in AlphaOS	1,259 human-labelled moments across 179 videos: the frozen clip-selection benchmark
~90 tables under 30 Row-Level Security policies in VerticalVoice's multi-tenant schema	~1,538 tests and 39 SQL migrations behind Clipper Tracking
10-step onboarding wizard and 10-step policy pipeline: a new voice tenant is live in 5 to 10 minutes	9 deterministic QC gates before any carousel leaves the building

🧠 How I build

1. Spec it like a contract        -> invariants first, features second
2. Test it like an adversary      -> golden corpora, eval harnesses, adversarial QA agents
3. Ship it like it's forever      -> idempotency keys, circuit breakers, atomic claims
4. Verify it like a skeptic       -> render the PDF and LOOK at it; text-grep gives false passes
5. Keep the model on a leash      -> code, not the LLM, has the last word

Agentic engineering with Claude Code. Multi-agent dev workflows: spec first, then implementation, then adversarial review passes, with the written spec committed and audited clause by clause.
Verification before ship. Golden sets, blind eval modes, acceptance suites over HTTP, and deterministic gates a generous model cannot talk its way past.
VPS + Docker + Traefik ops. Standalone containers behind one Traefik instance with automatic TLS, health endpoints the orchestrator actually watches, one-command deploys and rollbacks, PM2 and GitHub Actions where they fit.
Supabase / Postgres everywhere. Row-Level Security for tenant isolation, pgvector for RAG, parameterised repositories with server-only guards, idempotent migrations, integer cents for money.

🗺 The ecosystem

flowchart LR
    subgraph Voice["📞 Voice SaaS"]
        VV[VerticalVoice] --> TW[Twilio ⇄ Ultravox]
    end
    subgraph Intel["🧠 Agency intelligence"]
        AOS[AlphaOS] --> HCL[Hook + Caption Lab]
        AOS --> CL[Carousel Lab]
        AOS --> LE[Lead Engine]
    end
    subgraph Edu["🎓 Education agents"]
        SW[SeatWise]
        CP[CohortPilot]
    end
    subgraph Fleet["🤖 Automation fleet"]
        N8N[36 n8n workflows] --> AP[Autopost / publishing]
    end
    Voice --> DB[(Supabase · Postgres · pgvector)]
    Intel --> DB
    Edu --> DB
    Fleet --> DB
    DB --> DASH[Ops dashboards · Next.js]

🛠 Arsenal

Frontend

Backend & data

AI

Ops

📡 Recent transmissions

🎉 Merged PR #1 in FakharEAli/verticalvoice-ai

🐍 Contribution graph

Work with me

_{Source code for any system available under NDA.}
_{⚡ This README maintains itself. GitHub Actions update it while I'm building.}

Navigation Menu