
Muhammad Azhar
Alauddin
I build end-to-end applications with AI at their core — combining frontend, backend, and machine learning to create intelligent user experiences.

Experience
Professional journey
Roles where I shipped full-stack products end to end — frontend, backend, AI features, and the cloud infrastructure behind them.
Senior Software Engineer
- Architected a strategic migration from AWS to a Proxmox-based on-premise VM environment, cutting annual operational overhead by 40%.
- Built a high-velocity CI/CD ecosystem — containerized legacy services with Docker and orchestrated deploys via Dokploy, reducing deployment latency 70% through automated webhook pipelines.
- Pivoted AI infrastructure from third-party APIs to a self-hosted LLM architecture with Ollama, eliminating subscription costs and hosting models locally for data privacy.
- Hardened network security by replacing legacy Nginx HTTPS with Cloudflare Tunnels — secure, zero-trust external access with no open firewall ports.
- Developed an AI-driven GRC platform automating regulatory gap analysis via a hybrid RAG pipeline (Milvus + BM25 + LLM reranking) with real-time progress streaming and a human-in-the-loop reviewer workflow.
- Built a distributed real-time task engine with Celery and Redis for high-concurrency document ingestion and background vector embedding.
- Engineered a centralized observability stack with OpenObserve to aggregate logs and traces, reducing MTTR across distributed microservices.
- Shipped an org-wide self-hosted AI PR-review pipeline using Qodo PR-Agent, a GitHub App, Dokploy, and Ollama (qwen2.5-coder:14b) at near-zero cost.
Software Engineer
- Sped up KR-Type1 IMO DCS and CII Excel processing by containerizing AWS Lambda on ECR with openpyxl — 60s to 6s (≈92% faster).
- Optimized the Fleet Graph API with multi-threaded parallel computation — 40s to 5s (≈87.5%), adding Redis caching to bring repeat requests to ~1s.
- Implemented an in-memory batching mechanism for OpenObserve APM logging using a dequeue system — up to 90% fewer HTTP requests with zero log loss via shutdown handlers.
- Achieved a 48% Docker image size reduction (1.85GB → 955MB) through multi-stage builds and migrating pip to uv.
- Built a unified observability and deployment pipeline with middleware error handling (OpenObserve APM/RUM) and CI/CD via Cloudflare CDN and S3/Amplify.
- Enhanced marine reporting with LLM-based parsing for ship reports, strengthening KR-Type1 and EU-ETS logic through arrival–departure pairing and leg extraction.
- Built a multi-layered code-quality pipeline — Lefthook pre-commit with Biome, Ruff, BasePyright, plus PR-level Knip, jscpd, and ast-grep for dead code and duplication.
- Contributed major Maritime SaaS features and AI Agents across frontend and backend with Django, FastAPI, React, and Vue.
Part-time Full-stack Developer
- Set up GitLab CI/CD pipelines to automate deployment for the events and deals listing platform.
- Managed and optimized GCP infrastructure (Cloud Storage, Cloud SQL, Compute Engine), deployed frontends via Firebase Hosting, and resolved server downtime.
- Redesigned backend architecture and user flow to allow events and deals enrollment without login or registration.
- Supervised intern engineers with technical guidance and hands-on support.
- Managed 4 repositories (1 BE, 3 FE) for updates, bug tracking, and stability.
Full-stack Developer Intern
- Built Vue.js frontend features — notifications, newsletters, ship-spec and fuel-oil modals, SideNav overhaul, V-Select migration, account deletion, and a foldable chatbot.
- Improved backend performance by reducing query calls, adding backend pagination, and streamlining Django API interactions.
- Optimized Holt API interactions for EUA price comparisons using Redis caching.
- Implemented soft delete for critical entities — users, ships, and companies.
- Refined vessel-tracking maps with adaptive zoom and bounding-box calculations.
Part-Time Software Engineer
- Built responsive, mobile-friendly interfaces from mockups using React.
- Integrated APIs to enhance frontend functionality and data retrieval.
- Leveraged state management to share application state across components.
- Participated in Agile processes — sprint planning, stand-ups, and retrospectives.
Skills
The stack behind
production systems
The languages, frameworks, and infrastructure I use to ship and scale.
Frontend
Backend
Cloud & DevOps
Tools & Technologies
Projects
Featured projects
A selection of what I've built — production features, side projects, and experiments.






Activity
Building, most days
A year of commits, reviews, and side projects — straight from GitHub.
2,408 contributions in the last year
View GitHubWriting
Notes and reflections
What I'm learning as I build — engineering practice, career journey, life hacks.
Web3 Is Not Always About Crypto
Paying for overseas services like AI subscriptions with crypto is often far cheaper and faster than a bank and SWIFT — here's the flow I use.
Why Companies Run Local LLMs On-Premise
Many companies run local models on their own servers instead of the OpenAI or Anthropic API — driven by privacy, security, and deliberately layered releases.
The Engineering Skill AI Can't Replace
Anyone can build a product with AI now. The skill that keeps getting more valuable is knowing how to design a system the right way.
Don't Paste Secrets Into AI Prompts
Your prompts and AI responses can be captured by LLM observability tools like Langfuse. Keep API keys, tokens, and credentials out of them.
Contact
Let's build
something
Have a project, role, or idea in mind? Send a note and let's bring it to life with the right technology.