UK · Open to work

David
Tandoh

Backend & Cloud InfrastructureData & AI/ML PlatformsAgentic AI SystemsFinTech & Payments

I engineer resilient, intelligent systems for the financial sector — and I build agentic products end-to-end, from a live travel-tech marketplace to autonomous research and fraud-intelligence engines.

View selected workExplore my journey

00 — The human

Engineering at the seam of finance & AI

I’ve spent the better part of a decade close to money — first the card switches and payment rails that move it across Africa, then the cloud platforms and ML systems that decide whether to trust a transaction in the UK.

These days I build agentic systems end-to-end: a live travel-tech marketplace, an autonomous equity-research engine, a real-time fraud platform, and an always-on agent that runs a whole crew of coding agents. I care about systems that are resilient, auditable and honest about their own uncertainty — because in finance, the cost of being confidently wrong is real.

Resilient by default

Money-moving systems can’t fail quietly. I design for failure, observability and recovery first.

Intelligence with receipts

AI should explain itself. I favour signals and decisions a human can audit and trust — never a model marking its own homework.

Lift others up

I mentor underrepresented talent and helped build an offshore engineering team in Ghana — engineering is a team sport.

/ Now

What I’m focused on right now

  • 01Scaling GuideX — our travel-tech experiences marketplace for Africa, live on iOS & Android.
  • 02Building Bosun — a private, always-on chief-of-staff agent that runs a crew of coding agents.
  • 03Building DXJ Signal Engine — a weekly research agent over high-growth US stocks.
  • 04Shipping LLM-assisted fraud analytics for a UK retail bank at bigspark.
  • 05Mentoring young people in tech with Kocha Mentors CIC.

01 — Selected Work

Systems that move money & make decisions

From autonomous trading agents to bank-grade fraud platforms — a few things I’ve designed, built and shipped.

01 / Featured

GuideX

Travel-tech marketplace · Founder

Live2024 — present· Founder & Technical Lead

A travel-tech experiences marketplace — “where the world books Africa.” Travellers book tours, food, culture and nightlife with verified local hosts, starting in Ghana. Native iOS & Android apps, a web app, and a payments + KYC backend — with an autonomous AI SRE estate being built on top for reliability.

  • Founded the company and lead engineering across mobile, web and backend.
  • Shipped native iOS & Android apps — live on the App Store and Google Play — for discovery, booking, in-app chat, QR-code tickets and payments.
  • Built a two-sided marketplace with host KYC/KYB (SmileID) and GHS payments via Paystack (multi-currency price display; multi-currency settlement planned).
  • Designed an autonomous AI SRE estate: Sentry/GCP alerts route via Pub/Sub to LLM “flow specialist” agents (triage / payments / KYC) with a kill-switch, writing incident files and Slack threads.
Live
iOS · Android · Web
Africa
Market
FlutterNext.jsTypeScriptKotlinSpring BootFirebaseGoogle CloudPaystackSmileIDLangGraph
Live siteApp StoreGoogle Play
02 / Featured

Bosun

Agentic orchestration

Building2026· Creator & Engineer

A private, always-on “chief-of-staff” agent. Instead of juggling a dozen agent terminals, you talk to one thing — it dispatches a crew of coding agents across your projects from Kanban boards, gates risky commands to your phone for approval, and reports back.

  • A deterministic 24/7 outer loop watches agents and dispatches ready work — plain code, not an LLM, so autonomy stays in the inner loops.
  • A tiered, fail-closed approval gate: routine work runs free, but force-push, deploy, rm -r and secrets pause for a tap-to-approve on the phone.
  • Per-project Kanban boards as durable state — a crashed orchestrator rebuilds from card status; only “ready” cards can ever be claimed.
  • Real-time cost telemetry born from a 35M-token burn post-mortem: live spend monitoring, a cheaper brain, and a hard daily cap on the API key.
Always-on
Orchestrator
Multi-agent
Crew
PythonOmnigentHermesLLM AgentslaunchdKanbanCEL

A write-up is coming to the Writing section.

03 / Featured

DXJ Signal Engine

AI equity research

Building2026· Creator & Engineer

A weekly “research-not-advice” US stock signal. A deterministic algorithm scores high-growth names from price momentum and US House disclosed buys; a “Hermes” deep-research agent then writes cited bull/bear → judge dossiers on the top flags — with a CI-enforced firewall so the LLM can never touch the score.

  • Deterministic signal score: price momentum (0–60) + a US House disclosed-buy overlay (0–40), tiered High / Watch / Low.
  • A “Hermes” deep-research agent runs a brief → decompose → retrieve → bull/bear → judge loop, grounded read-only on point-in-time data.
  • A hard “purity” invariant — LLM output is context only and can never reach the score — enforced by a CI firewall.
  • A validity-first eval pyramid: golden tests, data-quality gates, a regime-split backtest gate, and point-in-time look-ahead invariants.
Weekly
Signal run
House × Momentum
Signal
PythonFastAPINext.jsPostgreSQLDeepSeekClaudeGitHub Actions
04 / Featured

Cygnal

Fraud intelligence platform

Building2026· Creator & Sole Engineer

A real-time fraud intelligence platform for African banks and fintechs. It ingests payments over ISO 8583 and REST (ISO 20022 planned), scores each transaction with deterministic rules (ALLOW / REVIEW / BLOCK) against a sub-25ms latency target, and records every decision in a hash-chained audit trail built for central-bank examiners.

  • A protocol-agnostic scoring pipeline (adapters → normalise → enrich → rules → decision → audit), engineered to a sub-25ms P99 target; ONNX ML planned for Phase 2.
  • An in-progress “Spec Translator” designed to turn proprietary switch specs (Postilion, SmartVista) into adapter configs via an evidence-first, AI-assisted pipeline.
  • Per-tenant SHA-256 hash-chained, append-only audit trail for examiner-grade compliance.
  • One codebase, two deployments: on-prem single-JAR for banks, multi-tenant cloud SaaS for fintechs.
Sub-25ms
P99 target
ISO 8583 / REST
Rails
KotlinSpring BootWebFluxReactPostgreSQLRedisISO 8583Kubernetes

Cynthia

Prototype

Multi-model engineering workflow

A governed multi-model engineering workflow — an agent pack that runs inside Omnigent (Databricks’ open-source agent control-plane). It orchestrates Claude as planner/reviewer and Codex as the sole writer under strict role governance, human approval gates, and external completion checks rather than model self-report.

PythonOmnigentMCPClaudeCodex

02 — The Journey

A decade engineering for finance

From card switches in Accra to AI-driven risk platforms in the UK — each step compounding toward intelligent financial systems. Every milestone is tied to where the industry is heading.

Jun 2021 — PresentUK

Senior Engineer

Bigspark — UK retail banking client

Leading platform development, fraud analytics and offshore team expansion for a major UK retail bank.

  • Designed a signal-driven risk pipeline with LLM-assisted classification for fraud investigation.
  • Built ML-driven fraud detection inside UK banking regulation, reducing false positives by ~35% and unlocking £2.5m forecast impact.
  • Engineered a multi-tenant data platform on AWS with Scala, Spark and EMR for self-service data science.
  • Architected a HA ingestion gateway (Apigee, Kafka, Spring Boot) feeding Snowflake securely.
  • Helped establish and mentor an offshore engineering team in Ghana, sustaining delivery quality.
Industry relevance

Sits at the intersection of regulated FinTech, cloud-native data engineering and applied AI — the exact blend driving banking’s shift to intelligent, automated risk.

AI/MLFraudAWSKafkaScalaLeadership
Jan 2021Leicester, UK

MSc, Software Engineering for Financial Services

University of Leicester

Master’s focused on engineering for capital markets and financial systems.

  • Time-series data processing, real-time ingestion and Efficient Portfolio Theory.
  • Hands-on with Bloomberg Terminal and market-data workflows.
  • Awarded Best MSc Student in Software Engineering for Financial Services (2020).
Industry relevance

Formal grounding in quant finance and market-data systems — the foundation under the DXJ Signal Engine.

Market DataQuantTrading Systems
Jul 2018 — Sep 2019Ridge, Accra, Ghana

Card Systems Engineer

EProcess International (Ecobank)

Owned critical card-payment systems and scheme integrations across a pan-African bank.

  • Managed and supported Postilion Realtime and Postilion Office card systems.
  • Led integrations of National Payment Switches with the bank’s core switch.
  • Technical liaison for VISA & MasterCard, ensuring seamless scheme integration.
  • Redesigned incident management and escalation, cutting response times by ~50%.
Industry relevance

Deep payments-rails expertise (ISO 8583, switching, card schemes) — the operational reality behind modern FinTech.

PaymentsVISA/MastercardPostilionISO 8583
Aug 2016 — Jul 2018Labone, Accra, Ghana

EFT Software Engineer

Electronic Funds Technology Corporation

Engineered card-payment integrations powering transactions across multiple African countries.

  • Built custom integrations on the Postilion Realtime Framework with the Java SDK.
  • Engineered Postilion Office plug-ins that automated reconciliation and lifted operational productivity by ~25%.
Industry relevance

Where it started: low-level payments engineering and the discipline of money-moving systems that cannot fail.

JavaPaymentsIntegration
Jun 2016Berekuso, Ghana

BSc, Computer Science

Ashesi University

Computer Science degree from one of Africa’s leading universities.

  • Foundations in software engineering, algorithms and systems.
Industry relevance

The starting line — rigorous CS fundamentals that scale from card switches to AI agents.

Computer ScienceFoundations

03 — Capabilities

The toolkit

Spanning cloud-native infrastructure, data engineering at scale, applied AI, and the payments rails underneath it all.

Cloud & DevOps

01
AWSAzureGCPDockerKubernetesTerraformGitHub ActionsGitLab CITeamCityApigeeTykSonarQube

AI, LLMs & Agents

02
OpenAI APIsLangChainLangGraphCrewAIRAGFine-TuningVector StoresEmbeddingsFunction CallingMCPClaude Code / Cursor

Data & Messaging

03
Apache KafkaApache SparkSnowflakeDatabricksStreamSetsEMRMongoDBPostgreSQLData LakehousesAirflow

Languages & Frameworks

04
JavaKotlinSpring BootPythonFlaskScalaGoC#ReactNext.jsFlutterRESTGraphQL

FinTech & Payments

05
ISO 8583ISO 20022Payment SwitchesFraud DetectionPCI DSSPostilion

Quality & Practices

06
TDD / BDDJestKotestJUnitMockitoTestcontainersAgileGit

04 — Recognition

Certifications & awards

Microsoft Certified: Azure Fundamentals

Microsoft

Databricks Certified Data Engineer Associate

Databricks

Building Visionary Partnerships

bigspark · 2023

Best MSc Student — Software Engineering for Financial Services

University of Leicester · 2020

05 — Let’s build

Have a hard problem in finance or AI? Let’s talk.

Open to senior engineering roles, AI/ML platform work and serious collaborations. The fastest way to reach me is a direct email.