Skip to content
Available now · Open to software engineering roles

David Manufor

Backend Engineer solving real world problems

Software Engineer · Lagos, Nigeria (GMT+1) · Remote

Engineering scalable systems and AI built for real impact: reliable services, clean architecture, and software that keeps working as it grows.

  • 4 yrsproduction experience
  • 20+client engagements
  • ∞cups of coffee
01

About

I'm a software engineer with four years of production experience across fintech, e-commerce and healthtech, focused on systems that stay reliable under failure and scale. I build REST and event-driven services in Python and Node.js on PostgreSQL, Redis, AWS and GCP.

Recent work includes a cross-border payment engine that settles three fiat currencies against an on-chain token, and a multi-tenant platform that tracks clause-level changes across contracts and privacy policies. I'm also fluent in AI-assisted development with Claude Code and Codex, using MCP servers, skills, plugins, subagents and deliberate context management.

I believe in constant growth, in code and in thought. I spend a lot of time building systems, breaking them, and writing about what the process teaches me over on my blog.

More about me on Human Readable

Off duty

  • CoffeePowers most commits
  • Formula 1Proudly cheering on McLaren
  • Table tennisFast reflexes, slow serves
  • AnimeAlways one more episode
  • Video gamesResearch into game systems
  • BooksLife and technology, lately
02

Experience

Astra Development Studio

Software Engineer

2022 – PresentRemote
20+ engagements deliveredFintech · e-commerce · healthtech
TypeScriptNode.jsExpress.jsPythonPostgreSQLStripePayPalReactNext.js

Registered consultancy delivering contract engineering to startups in fintech, e-commerce and healthtech. 20+ engagements delivered alongside full-time studies.

  • ▸Built end-to-end checkout and payments for a UK e-commerce client, integrating Stripe and PayPal with idempotency keys, automatic retries and webhook reconciliation to prevent duplicate charges and lost orders on dropped connections or repeated provider callbacks.
  • ▸Designed the payment backend and frontend interfaces for a startup fintech, covering mobile-money and card flows across third-party providers.
  • ▸Built a booking and scheduling platform for a Kenyan therapy provider with an accessibility-focused frontend, coordinating delivery across a 10-practitioner remote team.
  • ▸Shipped multiple custom e-commerce platforms against headless CMS backends, owning architecture, implementation and deployment.
⎇ internship/smartpay

SmartPay

Software Engineering Intern

Feb 2024 – Sep 2024Remote
3 fiat currencies + on-chain settlement13-person engineering org
GoExpress.jsNode.jsTypeScriptPostgreSQLRedisNext.js

Fintech startup building a hybrid fiat-cryptocurrency wallet for cross-border payments; 13-person engineering organisation.

  • ▸Co-built the cross-border payment engine in Go and Express, implementing multi-currency ledgers across three fiat currencies and the platform's on-chain token to settle fiat-to-crypto transfers.
  • ▸Established Next.js architectures and a reusable component library adopted by the three-person frontend squad as the wallet's design-system foundation.
  • ▸Co-authored design documents covering ledger structure, settlement paths and failure handling ahead of implementation.
  • ▸Reviewed code across backend services and frontend features as part of the team's review rotation.
03

Projects

ClauSync AI

Multi-tenant contract and policy change monitoring

Demo
ClauSync AI — application screenshot

A multi-tenant platform that tracks clause-level changes across contracts and privacy policies. A 12-service Turborepo monorepo moves scraping, diffing and clause analysis off the request path onto independently deployed Pub/Sub workers, so a single failing stage cannot halt processing and each stage scales on its own.

  • ▸Architected a 12-service Turborepo monorepo — REST API, web application, marketing site and nine independently deployed Pub/Sub workers — moving a nine-minute scrape, diff and clause-analysis pipeline off the public request path so a single failing stage cannot halt processing and each stage scales independently.
  • ▸Enforced tenant isolation in PostgreSQL through row-level security and a Prisma $allModels extension that applies the tenant predicate automatically, eliminating cross-tenant exposure from omitted WHERE clauses.
  • ▸Built the retrieval-augmented generation pipeline on pgvector and Vertex AI embeddings, grounding clause extraction in each tenant's own policy baseline, with confidence scoring and fallbacks for malformed model output.
  • ▸Provisioned 55 GCP resources as modular Terraform and moved schema migrations out of container startup into a Terraform-managed Cloud Run Job, eliminating cold-start races on deploy.
  • ▸Maintained 134 automated tests across TypeScript and Python services, alongside a product requirements document, architecture design document and API specification.
  • 12-service Turborepo monorepo
  • 9 independently deployed Pub/Sub workers
  • 55 GCP resources as Terraform
  • 134 automated tests
TypeScriptPythonGCPPostgreSQLpgvectorPrisma

BlackBridge

Zero-knowledge encrypted file transfer

Retired
BlackBridge — application screenshot

An end-to-end encrypted file transfer and file-request platform. Files are encrypted with OpenPGP in the browser before upload, authentication uses Secure Remote Password so the server never receives the password or a hash of it, and a Web Worker pool streams ciphertext in chunks to support multi-gigabyte transfers without exhausting browser memory.

  • ▸Implemented client-side OpenPGP encryption across a Web Worker pool sized to available CPU cores, streaming ciphertext in 64 KB chunks to support files up to 10 GB without blocking the main thread or exhausting browser memory.
  • ▸Implemented Secure Remote Password authentication so the server receives only a cryptographic proof of the password rather than the password or a hash of it, and added PGP public-key sharing so file keys never reach the server.
  • ▸Implemented encrypted peer-to-peer transfers using WebRTC and a custom signaling server.
  • ▸Built a 101-endpoint Express and Prisma API across 14 data models, hardened with Redis-backed rate limiting that holds across instances and short-lived signed CloudFront URLs for download authorisation.
  • ▸Provisioned 41 AWS resources as Terraform and validated authentication and transfer flows with Playwright end-to-end tests in CI.
  • ~500 users at peak
  • 101 endpoints across 14 data models
  • 41 AWS resources as Terraform
  • 10 GB transfers in 64 KB chunks
TypeScriptNode.jsExpress.jsPrismaPostgreSQLRedis

ViralCue

Real-time AI co-pilot for live streamers

In Development
ViralCue — application screenshot

A five-service system that ingests live stream audio, scores segments for virality with an LLM, and clips and transcodes the results to vertical video. An Express and TypeScript API gateway fronts Python services, connected asynchronously over Pub/Sub and Kinesis.

  • ▸Built a five-service microservices system with an Express and TypeScript API gateway (REST and WebSocket) fronting Python services for audio ingestion, AI analysis and video processing.
  • ▸Connected services asynchronously over Pub/Sub and Kinesis, running the stream monitor on GKE and clip and transcode jobs on Cloud Run.
  • ▸Integrated Twitch, Deepgram, TikTok and X APIs, and authored the deployment runbook, API documentation and GCP IAM setup guide.
  • 5-service microservices system
  • REST + WebSocket API gateway
  • GKE + Cloud Run workloads
  • Async over Pub/Sub and Kinesis
TypeScriptPythonGCPAWSExpress.jsWebSockets

prompt-template-manager

Python library for versioned prompt templates

Maintained
prompt-template-manager — application screenshot

An open-source Python library for composing and versioning reusable LLM prompt templates. Prompts live in YAML with per-model variants, recursive includes and environment substitution, backed by a CLI that validates a whole prompt directory. Published on PyPI under MIT.

  • ▸Published and maintain an MIT-licensed library for composing and versioning reusable prompt templates, through four releases under semantic versioning with 85% test coverage and support for Python 3.9+.
  • ▸Designed a YAML prompt registry that resolves model-tagged variants (gpt-4, claude-3, llama-3) from a single key, so swapping providers never means rewriting call sites.
  • ▸Implemented recursive template composition via {include}, environment substitution via {env:VAR}, and an LRU resolution cache that keeps deeply nested templates constant-time after first access.
  • ▸Shipped a ptm CLI whose validate command catches syntax errors, type errors and circular includes across an entire prompt directory before they reach production.
  • Published on PyPI — v2.0.1
  • 4 releases under semantic versioning
  • 85% test coverage
  • MIT · Python 3.9+
PythonPyYAMLpytestCLIPyPIAnthropic SDK
03

Skills

Select a skill to highlight the projects above that use it.

intermediate proficient advanced

Backend

Cloud & Infrastructure

Data & AI

Languages

Testing & Practices

Frontend

05

Education

Strathmore University

BSc. Informatics and Computer Science

Coursework: Distributed Systems, Operating Systems, Compilers, Data Structures & Algorithms, Database Systems, Computer Networks, Artificial Intelligence & Machine Learning, Information Systems Security Design.

Graduated Aug 2026

Nairobi, Kenya

Available for opportunities

Let's work together.

I'm looking for a software engineering role on a team building reliable systems at scale. If that's what you're building, I'd love to hear about it.

Email me

davemanufor@gmail.com · typically replies within 24 hours

Writing

Human Readable

Notes on code, systems and clean architecture.