Hello World
I am Abdullah Chaudary
Senior Full-Stack Engineer | Cloud DevSecOps & Systems Architect | AI/ML
I build and scale complex AI systems, and the infrastructure that keeps them up 24/7 with 99.95% uptime. I handle full product ownership from design to deploy, end to end. Twelve years of it, with US, UK and EU teams. I have cut cloud spend by 45%, closed 80% of critical vulnerabilities inside two quarters, and led engineering teams of 20 across five products at once. Right now that means putting AI into production: fine-tuned self-hosted language models, RAG pipelines and agentic automation, all on infrastructure my clients own outright.
<why>
What that gets you
I have lifted on-time delivery to 95% and cut production defects by 30%, running roadmap, sprint planning and code review across multiple engineering squads. I have also led engineering teams as large as 20 across five concurrent products at once. Hire me for delivery and releases stop being a guess.
I run production infrastructure at 99.95% uptime, and I own security the same way: WAF, SAST/DAST scanning, container scanning and IAM hardening that closed 80% of critical vulnerabilities inside two quarters. If your infrastructure is unstable or nobody owns its security, that is the work I do first.
Client work is under NDA, so I cannot name names. I can point to the architecture instead: an EdTech platform I moved from AWS to self-hosted infrastructure at roughly 70% lower infrastructure cost with zero downtime through the cutover, and a FinTech platform that shipped from a technical proposal I wrote myself.
If any of that matches a problem on your roadmap, I take on new engineering engagements and technical leadership roles.
Anything worth running in production is worth being able to rebuild from scratch.
The interesting part of a system is never the happy path. It is what it does when one piece fails.
Anything worth running in production is worth being able to rebuild from scratch.
The interesting part of a system is never the happy path. It is what it does when one piece fails.
Check out my
awesome work
Latest writing
Read the blogPostgres RLS Unique Constraint Leak: Hidden Rows Collide
A Postgres unique constraint ignores row-level security, so a tenant that cannot SELECT a row can still find it by INSERT. Scoping the key to UNIQUE (tenant_id, email) closes it.
Self-Hosted LLM for Sensitive Data: Hosting Is Not Security
A self-hosted LLM for sensitive data is not secure just because it is on-prem. vLLM's --api-key skipped /invocations and /metrics in my test, and an nginx allowlist closed both.
n8n Agent Approval Payload Hash: Bind the Yes to the Bytes
An n8n agent approval payload hash binds the reviewer's yes to the exact bytes: the gateway publishes only a payload whose SHA-256 was approved, and refused a regenerated draft in a scripted test.
Hybrid Cloud for Regulated Workloads: Not All On-Prem
Hybrid cloud for regulated workloads, from client engagements in health, pharma and education: most services moved on-prem, while model training and some backups stayed in the cloud.
Local LLM vs Cloud: Idle RAM Is Not Waste
Sam Altman calls RAM in personal computers wasteful at 5% use and wants maximum cloud utilization. In local LLM vs cloud, utilization is the landlord's metric.