Skip to content

●  AI consulting & implementation · Dubai

AI consulting and implementation that survives production.

Most AI work starts with the model. I start with how your business actually runs, advise on where AI and architecture genuinely serve it, then implement the systems myself: built to hold up under real load, real data and real regulators.

  • AI Consulting
  • Implementation
  • Technical Advisory

16 years · a limited number of assessments each month

i.Thesis

Great AI begins with a clear read of the business.

I start by finding where AI genuinely serves your operation: the places it compounds, the work it lifts, the decisions it sharpens. Then I build exactly that. The clarity is the work; the build follows from it.

ii.Method

How every engagement runs.

01

Read the business

Where AI genuinely serves the operation, and where it would only add noise. Written down before anything is built.

Observations
02

Architect

A design that fits the systems you already run, with security, data handling and cost decided up front.

Structure
03

Build in production

Working software against real data early, not a demo that has to be rebuilt later.

Form
04

Prove it

Measured against written success criteria agreed at the start, with evaluations you can rerun.

Measure
05

Hand over

You own the code and IP. Documentation and a runbook mean your team can run it without me.

Yours

iii.Selected commissions

Work that endures.

  • Engaged by EY on an internal AI workflow platform for software delivery, which EY is productising into its client offering. The work started as advice: as design authority I ran discovery with business and security stakeholders, produced the architecture views, threat model and non-functional requirements, chaired design reviews against cybersecurity and cloud guardrails, and led the build-versus-buy assessment for the LLM layer, recommending a provider-agnostic orchestration framework over a single-vendor SDK after a proof of concept. I then architected the platform and led its engineering team: agentic workflows that automate threat modelling, compliance, retrieval-grounded documentation, planning, code generation, security scanning and reviews; a trust-tier model that holds high-risk agent actions for verified human sign-off; an audit trail recording every agent action, tool call and prompt; and tenant isolation built into the data layer.

    Enterprise AI platform · AI workflow · Agentic AI governance · Design authority
  • Engaged through EY. The client was carrying heavy quality-assurance overhead and an enterprise design system stretched across many frameworks. I targeted the QA load with AI agents, a SonarQube remediation agent and an automated code review agent, and drove adoption across client teams by exposing the framework through a Model Context Protocol (MCP) server. I architected an end-to-end and integration testing framework whose specs are generated automatically through that MCP server. As design authority for the enterprise design system, I set the reference architecture and release standards for its consuming applications, led its migration to StencilJS with native Angular and React wrappers, built the Azure DevOps pipelines behind it, and led a distributed team across several countries.

    AI agents · MCP · Design system authority · Test automation
  • For Fidelity Investments' Ireland-based operation, I led the design and development of key components of a new application that lets portfolio managers visualise high-volume market data at speed, built in Angular and D3.js. I architected a shared Nx component library that cut development redundancy by about 20%, set up Cypress end-to-end testing with 80% unit test coverage, and mentored developers through regular code reviews.

    Data visualisation · Angular + D3.js · Nx component library · Test automation
  • A six-week buy-side technology due diligence for Roland Berger on a single acquisition target, which stays confidential. I led the architecture, code and SDLC, and AI workstreams: I challenged the seller's answers to the information requests, turned a set of apparent control gaps into one costed decision the deal team could carry into the valuation model, and tested the target's AI claims against evidence before valuation.

    Technical due diligence · M&A · AI assessment · Confidential
  • I am the fractional CTO for Interview Strategist, a platform that turns a professional's experience into stakeholder-aware CVs, LinkedIn positioning and interview preparation, all drawn from one structured evidence library instead of being improvised for each application. I wrote the full design suite before the build (requirements, architecture, threat model, NFRs and 35+ architecture decision records) and build the platform to it, including its AI agent pipelines, with personal data stripped before every model call. The product is pre-launch.

    Fractional CTO · Pre-launch · AI product · Architecture
  • The department ran many applications without a shared design language, and wanted to know where generative AI could help its staff. I took over the architecture of its bespoke design system, acted as the client's technical point of contact for the 12 applications adopting it, and advised design and business stakeholders on high-stakes UX decisions. In a later engagement I set the frontend architecture and design system strategy for the department's Angular multi-application estate, led a team of 6 engineers with 90% unit test coverage, and shaped a GenAI use case: a retrieval-augmented (RAG) proof of concept giving staff grounded answers from the department's Confluence knowledge.

    Design system · Frontend architecture · RAG proof of concept · Module federation
  • A direct-to-consumer beauty brand needed to sell in the UAE without handing checkout to a third-party domain or treating customer data as an afterthought. I advised on the commerce architecture and the data obligations first, then built a headless storefront on Next.js and Medusa with checkout kept on the brand domain, verified payment webhooks, authenticated admin and operations endpoints, and data handling designed around the UAE data-protection law (PDPL), including consent, retention, and erasure.

    Headless commerce · On-domain checkout · UAE PDPL · Next.js + Medusa
  • I worked in a dual role as Scrum Master and Senior Angular Developer for one of Ireland's largest retail banks, running agile delivery across complex banking applications while staying hands-on in the code.

    Angular · Scrum · CSM · Banking

iv.Services

Craft over commodity.

From AI strategy to production. First, where AI genuinely serves the operation; then agents, retrieval over your own knowledge, and automation wired into the systems you already run. Built with evaluation, cost control, and audit trails from day one, for teams in the UAE and Ireland.

Read the full service

  • AI strategy and use-case selection
  • Where AI serves the operation, written down first
  • Retrieval over company knowledge (RAG)
  • Proving it against written success criteria
  • Evaluation, LLM cost and context control

Agents are a decision before they are a build. I start with which work an agent should take, which stays with a person, and what a wrong answer would cost. Then I build it: agents and multi-agent workflows that run inside the systems you already have, with every action recorded.

Read the full service

  • Agent and multi-agent workflow design
  • Automation inside the systems already in place
  • Model Context Protocol (MCP) integrations
  • Human sign-off for high-risk actions
  • An audit trail of every action, tool call and prompt
  • Agent evaluation and running-cost control

Architecture consulting that carries through to the build. A senior read of what you run today, a target architecture the team can follow, and software designed to outlast its first brief, with design systems treated as part of the architecture.

Read the full service

  • Enterprise architecture consulting and reviews
  • Platform and solution architecture
  • Integration and API architecture
  • Security, data handling and cost decided up front
  • A target architecture the team can follow
  • Legacy platform modernisation

Web applications and sites, engineered to a reference architecture rather than assembled. I settle the decisions first: release standards, and build versus buy for the component layer. Then I build it, design the interface that serves it, and set the design system that keeps your applications consistent.

Read the full service

  • Web application and site engineering
  • Frontend architecture for estates of many applications
  • Interface design that serves the build
  • Design-system strategy, architecture and governance
  • Accessibility-first component libraries
  • Release standards and the pipelines behind them

For web and mobile products, from the first strategic question to release. I start with advice: what the product must achieve, the technology strategy, and the platform choice between native, hybrid and web. Then I lead the build, with the interface designed as part of it, test and release it, and hand it over with the documentation and pipelines your team needs to own it.

Read the full service

  • Product and technology strategy
  • Architecture and platform choice: native, hybrid or web
  • Mobile application build
  • Testing, release and CI/CD
  • Handover, documentation and team enablement

For leaders who need a senior read before they commit, and someone accountable after. Where AI genuinely serves the operation, what to build or buy, whether the architecture will carry it, and a fractional CTO when the team needs one.

Read the full service

  • AI readiness assessments
  • Build-vs-buy and technology decisions, including the LLM layer
  • Architecture and code reviews
  • Technical due diligence
  • Fractional CTO for early-stage founders
  • Direction for in-house engineering teams

Headless e-commerce for brands selling in the Emirates and in Ireland, from platform advice to launch. On-domain checkout, regional payments, and data protection under the UAE PDPL and the EU GDPR treated as requirements, not afterthoughts.

Read the full service

  • E-commerce platform advice and architecture
  • Headless storefronts (Next.js, Medusa)
  • On-domain checkout and regional payments
  • UAE PDPL and EU GDPR-aware data handling
  • Performance and conversion-ready landing pages
  • Catalogue, order, and fulfilment integrations
Portrait of Abhijit Chatterjee Chattopadhyay

v.The architect

Abhijit Chatterjee Chattopadhyay

Sixteen years of building enterprise platforms across Ireland and Europe, distilled into a practice where engineering rigour is the starting point for every commission.

Principal · Dubai

Who builds this
Before the commission

Begin with clarity.

Thirty minutes, free. I review your stack, you bring the challenge. No obligation. We pick one of the four below, I tell you where I would start, and what the work would take.

What I cover

  • AI product readiness: where your stack supports or blocks integration
  • Frontend and platform architecture: performance, scalability, and team velocity
  • Design system maturity: gap analysis and build-vs-buy framing
  • Technology strategy: decisions, pivots, and the cost of current choices

What you walk away with

  • A short written summary of what I would do first, and what that work would involve
  • A clear suggested next step, whether that involves me or not
Book the free assessment

I take a limited number of assessments each month · Video call · UAE · Ireland

The assessment is free. Anything that follows is quoted in writing before it starts.

~/thelogicatelier · practice.md

AI consulting and implementation that survives production.

Most AI work starts with the model. I start with how your business actually runs, advise on where AI and architecture genuinely serve it, then implement the systems myself: built to hold up under real load, real data and real regulators.

operation.field · live
x -- · y -- · value -.---point or tap to read
practice
consulting · build · advisory
experience
16 years
region
DXB · UTC+4 · Dublin +3h to +4h
availability
a limited number each month

// thesis

Great AI begins with a clear read of the business.

I start by finding where AI genuinely serves your operation. Then I build exactly that. The clarity is the work; the build follows from it.

approach.diff+3 −1
1−start with the model, then look for a use
1+start with how the business actually runs
2 find where AI genuinely serves it:
3+ the places it compounds
4+ the work it lifts, the decisions it sharpens
5 then build exactly that

01 · method

How every engagement runs.

  1. stage 01passed

    Read the business

    Where AI genuinely serves the operation, and where it would only add noise. Written down before anything is built.

    • assessment.md
  2. stage 02passed

    Architect

    A design that fits the systems you already run, with security, data handling and cost decided up front.

    • architecture.md
    • security.md
    • cost-model.md
  3. stage 03passed

    Build in production

    Working software against real data early, not a demo that has to be rebuilt later.

    • working software
    • real data
  4. stage 04passed

    Prove it

    Measured against written success criteria agreed at the start, with evaluations you can rerun.

    • success-criteria.md
    • evals/
  5. stage 05passed

    Hand over

    You own the code and IP. Documentation and a runbook mean your team can run it without me.

    • source + IP
    • docs/
    • runbook.md
› 5 of 5 stages passed. Handed over: code, IP, runbook.

02 · selected commissions

Work that endures.

8 of 8 rows
Selected commissions, sortable by column
Expand
EYInternal AI workflow platformAn internal AI workflow platform for EY, being productised for its clients.ai
Walgreens Boots Alliancedelivered via EYAI agents, test automation and design-system authority for Walgreens Boots Alliance.ai2025
Fidelity InvestmentsIrelandHigh-volume market data, visualised for portfolio managers at Fidelity Investments.architecture2023
Roland BergerGlobal strategy consultancyTechnical due diligence for Roland Berger.advisory2026
Interview StrategistPre-launch, in progressFractional CTO for a career decision-intelligence platform.advisory2026
National government departmentEU member stateDesign system, frontend architecture and a GenAI proof of concept for a government department.architecture2021–2025
Mysha BeautyUAE direct-to-consumer beauty & sun careA secure, on-domain commerce platform for a UAE beauty brand.commerce2026
National retail bankIrelandAgile delivery and Angular engineering for a national retail bank.delivery2020

03 · services

Craft over commodity.

ai-strategy-implementation.ymlyaml
name: AI Strategy & Implementation
summary: >
From AI strategy to production. First, where AI genuinely serves the operation; then agents, retrieval over your own knowledge, and automation wired into the systems you already run. Built with evaluation, cost control, and audit trails from day one, for teams in the UAE and Ireland.
includes:
- AI strategy and use-case selection
- Where AI serves the operation, written down first
- Retrieval over company knowledge (RAG)
- Proving it against written success criteria
- Evaluation, LLM cost and context control

04 · about

whoami

Portrait of Abhijit Chatterjee Chattopadhyay
principal.webp
$ whoamiexit 0
name
Abhijit Chatterjee Chattopadhyay
role
AI Consultant, Advisor and Implementation Architect
based
Dubai, UAE
experience
16 years · enterprise platforms across Ireland and Europe, now Dubai
focus
AI systems · frontend architecture · technical advisory
stack
Angular · TypeScript · Node.js · Python

Engineering rigour is the starting point for every commission. Who builds this ↗

05 · before the commission

Begin with clarity.

A limited number each month
requestvideo call · Dubai · Abu Dhabi
POST /consultation
{
  "duration": "30 minutes",
  "price": "free",
  "you_bring": "the challenge",
  "i_review": "your stack",
  "obligation": false,
  "you_get": "where to start"
}

covers

  • AI product readiness: where your stack supports or blocks integration
  • Frontend and platform architecture: performance, scalability, and team velocity
  • Design system maturity: gap analysis and build-vs-buy framing
  • Technology strategy: decisions, pivots, and the cost of current choices
response200 OK
// what you walk away with
  • A short written summary of what I would do first, and what that work would involve
  • A clear suggested next step, whether that involves me or not
Book the free assessment

The assessment is free. Anything that follows is quoted in writing before it starts.

Questions.

What does The Logic Atelier do?

The Logic Atelier is a practice based in Dubai, working with teams in the Emirates and in Ireland. It advises on where artificial intelligence and architecture genuinely serve a business, then builds the systems that deliver it. Seven services cover the range, from AI strategy and agents through enterprise architecture, web engineering and design systems, digital product and commerce, to senior advisory including fractional chief technology officer work. Start here sets out how that runs in practice.

Who will I be working with?

Me, from the first call to the handover. This is a practice rather than an agency, so the person you meet in the assessment is the one who writes the architecture and the code. There is no account manager in between and nobody junior working out of sight. Where a commission genuinely needs more hands than mine, I say so before it starts rather than quietly assembling a team, and the practice stays small by choice.

Do you advise, or do you build?

Both, and the sequence is the point. Written advice stands on its own, so a target architecture, a build-versus-buy recommendation or a diligence report is a finished piece of work you can act on with or without me. Where I also build, the person accountable for the recommendation is the one accountable for the result, which is the opposite of handing a strategy document to a delivery team and hoping it survives the translation.

Why do AI pilots fail to reach production, and what if ours should not?

Most often because the use case was chosen before the operation was understood, which is why the read comes first here and is written down. If the honest answer is that AI would only add noise, I say so in writing, and that is a finished piece of work rather than a failed one. The advisory page answers what happens when the recommendation is not to build.

vi.Contact

06 · contact

Talk to me.

Talk to me.

Production AI and software architecture. Based in Dubai, working with teams in the Emirates and Ireland.

I work Dubai hours, Monday to Friday, the same week you do. Dubai is three hours ahead of Dublin in summer and four in winter, so my afternoon is your morning.

I will never sell your details, for any reason, and I never send spam.

How I use your enquiry
  • Your enquiry comes to me, Abhijit Chatterjee Chattopadhyay of The Logic Atelier. The form is handled by Formspree, which passes it to my inbox and keeps a copy in my Formspree account.
  • I use it to read your enquiry and reply. If no work follows, I delete it within 12 months, and sooner whenever you ask.
  • If Google Analytics is running on your visit, it is told that an enquiry was sent, and nothing you typed.
  • To see or delete what I hold about you, email hello@thelogicatelier.com.
  • The full detail is on the privacy page.