Skip to content
Adeleke Olamiposi, smiling in a navy suit and red tie

Hi, I’m Posi.

I’m a software engineer and AWS Certified Solutions Architect. I build the backend and cloud systems behind products: APIs, payments and webhooks, queues, retrieval over your own data and agent workflows, with the React and Next.js interface on top. I take projects from a blank repo to real users, and the work below shows what I built on each.

Selected work

EdTechShipped

Law Angels UK

Co-founder and backend engineer

I built the backend: Django REST Framework APIs and PostgreSQL services for authentication, timed mock exams, Stripe checkout, subscription webhooks and paid-content access.

  • Stripe checkout and subscription webhooks behind monthly, quarterly and annual tiers
  • Subscription webhooks and paid-content access in the backend
  • Timed FLK1 and FLK2 mock exams with progress tracking
  • Django REST Framework
  • PostgreSQL
  • Stripe
Dev toolsIn progress

Docny

Founding engineer

I built the TipTap documentation editor end to end, co-built the GitHub-based documentation generation, and shipped the embedded AI assistant that answers reader questions from published docs.

  • TipTap editor wired to save APIs and document persistence
  • Documentation generated from a connected GitHub repository
  • Integrations with GitHub, Jira, Linear, Slack, Notion and Algolia
  • TipTap
  • GitHub
  • Algolia
  • LLM assistant

More work

School ERPShipped

Axis by Digenty

Full stack engineer

A Java 21 and Spring Boot backend with a Next.js client: MySQL through JPA, Flyway-versioned schema changes, Spring Security auth and a multi-stage Docker build.

  • Java 21
  • Spring Boot
  • Next.js
  • MySQL
  • Docker
MobilityShipped

DondaX Limited

Frontend developer (contract)

A multi-step order form and a rider management app for 200+ riders, with the frontend integrated against Django APIs and a tracking database.

  • React
  • TypeScript
  • Tailwind
  • Django REST APIs
CorporateShipped

The Hague Industries

Designer and developer

A fully static corporate site: sub-second loads, content structured around the firm's service lines, and an inbound contact pipeline for new briefs.

  • Static site
  • Contact pipeline
More on GitHub

Open source

Public repositories you can read, run and fork.

All repositories
Architecture diagram: a FastAPI service on EC2 calls Amazon Bedrock, reads documents and the vector index from S3, runs async jobs through SQS with a dead-letter queue and DynamoDB job state, and reports metrics to CloudWatch

Architecture

1 / 3

Posiislight/bedrock-rag-platform

Bedrock LLM and RAG platform

An LLM service on AWS: a FastAPI app on EC2 calls Amazon Bedrock, answers questions over documents in S3 with citations, and the whole stack is defined in CloudFormation.

  • Retrieval over S3 documents with Titan embeddings and cited answers
  • Async inference through SQS, with a dead-letter queue after 3 failed receives
  • CloudWatch metrics for latency, tokens, errors and cost per request
  • API key generated in Secrets Manager; the instance role can only invoke two named models
  • Python
  • FastAPI
  • Amazon Bedrock
  • CloudFormation
  • SQS
  • DynamoDB

Posiislight/agent-workspace

Agent workspace

Describe a change in plain English, pick a GitHub repo, and five stages take it to a real pull request: coding, testing, review, human approval, then commit.

  • Each task runs on its own agent VM, which sleeps while waiting for approval so idle time isn't billed
  • The tester runs the repo's real test suite and sends failures back to coding
  • Retries are bounded: 3 test failures or 2 rejected reviews stop the task for a human
  • Approve or reject from the UI or from the pull request on GitHub
  • Python
  • FastAPI
  • LangGraph
  • React
  • Postgres
  • Redis

Posiislight/gremlin

Gremlin

Browser-agent QA built at the OpenFire hackathon. An LLM drives Playwright through a web app, deterministic oracles detect the bugs, and each finding comes with a replay and a fix handoff for a coding agent.

  • Fastify REST API and an MCP server with 11 tools, with BullMQ workers and Postgres (Drizzle) behind them
  • The model navigates and deterministic oracles detect: crashes, console errors, failed requests. Each agent is bounded by steps, time and tokens, and stops after repeated no-ops
  • A Chromium-level request policy blocks payment hosts, off-site navigation and writes in read-only mode; credentials are placeholders the model never sees
  • Autofix opens a PR only after the tests are not weakened, the repro passes and a separate review pass approves
  • TypeScript
  • Playwright
  • Fastify
  • BullMQ
  • Postgres
  • Next.js

About

I work closest to the parts people never see, and own a project end to end: architecture, APIs, deployment and the UI on top.

I'm Posi. Most of my work is for founders and teams who need one person to take a product from a blank repo to real users, and still be around when it has to scale.

More and more of that work is AI. At Docny I shipped an assistant that answers readers’ questions from published documentation. At i4nnova I built an embeddings search over issue logs that cut average resolution time by 40%. I’m also an AWS Certified Solutions Architect, which is why my LLM services come with queues, metrics and cost tracking.

What I build

AI features and agents

Assistants that answer from your own documents, retrieval over your data, and agent workflows with tests, limits and human approval built in.

MVPs and SaaS platforms

From idea to production: auth, payments, dashboards and AI features. Everything a product needs to take real users on day one.

Platforms and integrations

Stripe billing, subscriptions, referral systems, admin panels and third-party APIs wired into one reliable system.

Have something that needs to hold up?

adelekeolamiposi@gmail.comClick to copy

Open to freelance projects

Projects, roles or a half-formed idea. I reply within one to two business days.