A Finnish proptech platform where an XGBoost model trained on historical apartment transactions turns an address into an instant valuation — and a valuation into a commission-based sale mandate. I own the model, the warehouse, the scrapers that feed it, the realtor and admin portals, and the WebSocket dashboards on top. Retrained and re-validated every three to four months against newly ingested sales.
Moin Akmal Khan — Senior Software Engineer

Senior Software Engineer
Python back-end · AI / ML systems · Full-stack web
I build the things other people got stuck on.
Taught myself to code at fifteen, starting with cyber-security and the parts of a computer nobody looks at. Seven years on, that's still the job.
↓
Moin Akmal Khan
- Lahore, Pakistan
- 2019
- English, Urdu, Punjabi, Hindi
- Sponsorship welcome
I work best at the seams: where a model meets a database, where a scraper meets a business that bills on its output, where an undocumented legacy system meets a deadline. I've owned an entire technical estate alone — web app, realtor portal, admin platform, warehouse, scrapers — and I've dropped into distributed teams on a daily ticket cadence without friction.
Nothing about that came from a curriculum. It came from wanting to know how the machine works, one layer further down, and then one more.
“Hand me the problem nobody else could crack. However difficult, it gets built.”✳
Years shipping production software
Countries whose teams I've built for
Engineer running an entire platform
Active users on platforms I've scaled

The replacement for the spreadsheet every QA team secretly still runs on: suites, execution runs and results in one platform, with AI-generated test cases. I built the administration portal end to end — tenants, users, audit logging — and shipped Bring Your Own Storage so customers can point the product at their own back-end.
Registering a chemical means transcribing toxicology studies into IUCLID by hand, for days. I built the RAG pipeline that ends that: study PDFs ingested into a Weaviate vector database, LLM retrieval populating structured IUCLID records directly. Shipped publicly as chemassist.ai.
A LinkedIn-shaped network built from the back end out: profiles, connections, feeds, posts, comments and reactions, with privacy controls, background jobs for notifications and analytics, and feed generation tuned for social-interaction volumes that punish naive queries.
Back-end APIs and business logic with server-side processing and layered caching, with endpoints designed around SEO-driven content delivery and structured data consumption — built in close contact with the front-end team rather than thrown over a wall.
Where the work happens.
Open source
live from the GitHub APIGold user can access 10 times a minute, Silver 5 times and Bronze 2 times. Unauthenticated users shouldn’t be able to access more than once a minute.
Bakery management system using Tkinter
A simple Django web application to manage long and short term tasks
Website build with React.js to view any GitHub profile
—
A simple Number Guessing mobile game using React Native
Blok.ai
Lead Developer — Sole Engineer
- Sole engineer for the company's entire technical estate — customer web app, realtor portal, admin platform, data warehouse and scraping infrastructure — across a large multi-repository codebase.
- Took over a completely undocumented legacy system after the previous engineering team was dissolved; reverse-engineered the architecture and restored a working delivery pipeline for features and fixes.
- Own the XGBoost apartment price-prediction model end to end, retraining and re-validating it every three to four months against newly ingested Finnish sales data.
- Build and maintain the automated scrapers and ETL processes feeding sold-property data into the warehouse, and act as the single technical point of contact for a commission-based business.
- Deliver real-time dashboards over WebSocket, with Redis caching and query optimisation keeping the platform responsive at scale.
TestFiesta
Software Engineer — Contract
- Shipped features and defect fixes on a SaaS test-management platform replacing spreadsheet-based test-case management for QA teams.
- Built the administration portal end to end, covering tenant and user management.
- Integrated AI services for automated test-case generation and validation, and implemented Bring Your Own Storage so customers can connect their own storage back-ends.
- Implemented data encoding/decoding handling, authentication and audit logging for enterprise QA teams on a daily ticket cadence inside a distributed team.
SilverBrain.ai
Full-Stack & AI Engineer — Contract
- Built IUCLID Assist — live as chemassist.ai — automating regulatory data entry into the IUCLID chemical-safety database used for chemical registration.
- Designed and implemented the retrieval-augmented generation pipeline: toxicology study PDFs into a Weaviate vector database, LLM retrieval populating structured IUCLID records.
- Designed scalable REST APIs for complex domain logic, data validation and visualisation requirements.
- Contributed to a second product initiative (Nexus), discontinued for competitive market reasons.
Linked Matrix
Senior Developer
- Core engineer on a paid SEO SaaS suite serving millions of active users and generating multi-million-dollar annual revenue.
- Developed and scaled advanced Django and Django REST Framework services supporting business-critical workloads.
- Optimised database schemas and complex queries, improving reliability and cutting response times; containerised services with Docker, ran workloads on Kubernetes, maintained CI/CD.
- Implemented logging, monitoring and error handling to improve stability, and drove code-quality practice on a large, long-lived production codebase.
Moadbus
Software Engineer — Banking Applications
- Developed and maintained banking application features in a regulated financial-services environment, working across secure back-end services and sensitive data handling.
Early-stage start-up
Software Engineer
- Built phishAnt — a Python/Tkinter desktop application and companion Chrome extension classifying URLs as phishing or legitimate with a machine-learning model.
- Developed an automated image-enhancement tool integrating the PerfectlyClear engine, and contributed to an in-house hospital management system.
Grouped the way I actually think about them, not as a wall of logos. Depth is concentrated in channels 00 to 02; the rest is working fluency, earned by shipping with it.
Languages
- Python
- JavaScript
- TypeScript
- SQL
- PL/SQL
Back-end
- Django
- Django REST Framework
- FastAPI
- Flask
- Node.js
- Express.js
- SQLAlchemy
- Celery
- Async API design
- Microservices
- Auth / RBAC
AI & Machine Learning
- XGBoost
- RAG architectures
- Weaviate
- LangChain
- LangGraph
- LlamaIndex
- OpenAI API
- Claude API
- Agentic AI
- Recommendation engines
Front-end
- React
- Next.js
- Redux
- Tailwind CSS
- Framer Motion
- Material UI
- Bootstrap
- HTML
- CSS
Data & Storage
- PostgreSQL
- Redis
- Query optimisation
- Indexing & pagination
- Data warehousing
- ETL pipelines
- Web scraping
- Oracle APEX
DevOps & Cloud
- Docker
- Kubernetes
- Helm
- Kafka
- CI/CD
- AWS
- DigitalOcean
- Heroku
- Linux administration
- Git
Testing & Tooling
- Pytest
- Unit & integration testing
- Selenium
- Jira
- Webpack
- Vite
- Stripe
Ways of working
- System design
- Debugging & performance
- Code review
- Distributed remote teams
- Direct client communication
Education
BS in Software Engineering
Virtual University of Pakistan
Intermediate in Computer Science
Standard College
Languages
- English
- Urdu
- Punjabi
- Hindi
Availability
- Remote worldwide
- Open to relocation
- Visa sponsorship welcome
Certifications
- 01Django for Everybody
- 02Complete JavaScript Development Bootcamp
Still learning
Electrical engineering
Self-directed and ongoing — embedded systems, computer architecture and low-level programming, because the layer under the one you work in is always worth knowing.
On file
- 15 September 2002
- Pakistani
- On request
Rabbit holes
- Operating systems
- Computer architecture
- Embedded systems
- Assembly-level programming
- DevOps
- Agentic AI
Let's build something hard.
Open to remote roles and relocation worldwide. Visa sponsorship welcome.
If it's the problem that's been sitting in the backlog because nobody could crack it — that's the one I want.