AI ENGINEER · TECHNICAL FOUNDER

Justina Ighodalo

I build AI systems for regulated industries, then stay with the customer until they work in production.

Founder-engineer with an enterprise risk and controls background, including at Shell. I have designed, built and deployed more than ten multi-tenant SaaS products across the UK, US and Nigeria, in governance and compliance, healthcare, construction, education and SME finance. I turn messy operational and regulatory requirements into data models, workflows and secure AI features, then run them in production.

Open to: AI Engineer. [LOCATION, e.g. UK-based, open to remote]

10+

products built and deployed

3

markets: UK, US, Nigeria

80+

Postgres tables under row-level security, audited

9

AI features routed through a private AWS Bedrock gateway

01 /

What I bring to a deployment

Discovery with the customer

A regulated financial institution evaluating a proof of concept asked for business continuity (BCP and BIA), DPIA, AI governance and internal audit modules. I scoped and shipped all four, and designed the AI path so prompts are not stored, shared with model providers or used for training.

Deploying into constrained environments

Private AI gateway on AWS (API Gateway, Lambda, Secrets Manager, Bedrock), with a target architecture for in-country hosting and a dedicated per-client model deployment where data residency rules require it.

Data migration and integration

Migrated about 892 historical member records into a client CRM. Built XLSX bulk import with row-level validation, preview before commit and round-trip export. Designing assisted migration from incumbent care systems.

Security and governance by design

Row-level security on every table, append-only audit logs enforced by database triggers, separation-of-duties rules in the data layer, and structured audits for cross-tenant access and data leaking through error messages.

AI with guardrails

Deterministic scoring that AI cannot override, generation grounded only in the selected regulatory framework, human approval before AI output becomes a record, and per-tenant usage metering with a clear fallback.

02 /

Work

More builds

LanedaGuard

Live, agent layer in design

UK GRC SaaS for SMEs with seven assessment tracks, Risk Chain, AI risk profile generation, policy lifecycle and governance structure modelling. AI calls moving to Bedrock in London (eu-west-2). Also designing LanedaGuard for Agents: a GRC tool layer AI agents can call to run assessments, controls and remediation end to end.

LDN Learning Academy (UK) and LDN International Academy (Nigeria)

Live

EdTech platforms, the UK one live with paying subscribers on Stripe, the Nigeria one on Flutterwave recurring billing. Designed AI curriculum spirals delivered as JSON and SQL (UK: 37 units, 148 topics, 406 hours. Nigeria: 38 units, 151 topics, 418 hours). Parent dashboards with weekly reports. GCSE revision for AQA and Edexcel built on a curated question bank rather than AI-generated questions, to avoid wrong answers.

Mudeme BizControl

Live

business management for Nigerian SMEs. Flutterwave recurring billing in Naira and USD, debtor reminder emails, an AI assistant on Claude Haiku with monthly caps, and email OTP as a second factor.

Church CRM (client delivery, UK)

Delivered

migrated about 892 historical member records, Kanban follow-up pipeline, two-tier admin roles and automated SMS through Twilio from edge functions.

Echo Legacy (public code)

Open source

family voice-story app in Next.js 14 (App Router) with Supabase (8 tables, RLS, storage), Whisper transcription and GPT summaries, Stripe subscriptions with webhooks, and shared vaults with invite codes and search.

Agent Platform Studio (prototype)

Prototype

an internal agent hub concept with configurable, skill-based agents, decision forums and a path-to-production process for taking an agent from idea to live. Runs locally against the Claude API.

Mudeme Partner Hub

Delivered

partner referral portal with a customer verification engine and commission tracking.

03 /

How I work

01

Embed

sit with the people doing the work and the people accountable for it.

02

Map

turn the regulation, process or pain point into entities, states and rules.

03

Build

ship a working slice quickly, then iterate with the customer in short cycles.

04

Harden

access control, audit trail, failure handling, cost limits and security review.

05

Hand over

walkthroughs written role by role, so the team can run it without me.

How I build

I use AI-native tools (Lovable, Claude Code) for speed on interface and boilerplate. I own the parts that decide whether a system survives production: the data model, access control, cloud architecture, integrations, security review and deployment.

04 /

Stack

Languages and frameworks

TypeScriptJavaScriptReactNext.jsNode.jsDenoPythonSQL

Data and backend

PostgreSQLSupabase (RLS, triggers, edge functions, auth, storage)REST APIswebhooks

Cloud

AWS BedrockLambdaAPI GatewaySecrets ManagerSESAmplifyVercel

AI

Claude (Sonnet, Haiku)OpenAI (Whisper, GPT)embeddingsgrounded generationhuman-in-the-loop designusage metering

Integrations

StripePaystackFlutterwaveTwiliohCaptcha

Tooling

GitHubClaude CodeLovable
05 /

Background

Enterprise risk and controls, including at Shell. [ADD ROLE TITLE AND YEARS]

GRC consulting for European clients: NIS2 compliance mapping for a client in Italy, covering 116 requirements mapped at requirement level, 270 control-to-application applicability decisions, and 105 remediation gaps reconciled one to one with control coverage.

Founder, Laneda Technology Ltd (UK) and Mudeme Technology (Nigeria): a product portfolio across GRC, care, construction, education and SME software.

Regulatory frameworks worked with: NDPA 2023, GAID 2025, CBN AML/CFT/CPF, NIS2, ISO 27001, ISO/IEC 42001, EU AI Act, NIST AI RMF.

Education and certification: [ADD: MSc Computer Science, institution, year]. CISM (in preparation).

06 /

Contact

Let's talk about your next deployment.