Explore my work ↗ / LinkedIn / Get in touch
I'm Abdullah Al Sami, a developer working across AI, machine learning, and ML research. I like understanding how models behave, testing ideas through experiments, and turning them into software people can use.
My work also spans Shopify, payment integrations, business dashboards, full-stack development, and DevOps — the interfaces, services, and infrastructure that turn an idea into a working system.
Currently focused on → ML reliability and evaluation, language models, AI tooling, and practical AI applications.
Refreshes every six hours via GitHub Actions. Activity covers the rolling contribution year; longest streak is within that window. Current streak uses UTC and allows today to be unfinished.
Explore the numbers & refresh status ↗
Contribution history · Recent public activity · Refresh workflow
Counts reflect what GitHub exposes to the workflow token. Commit contributions follow GitHub's contribution rules; they are not a count of every commit on every branch. Languages measure code bytes in the 30 most recently pushed public, non-fork, non-archived repositories, excluding this profile. They do not measure proficiency or time spent coding.
|
Language models, AI applications, model routing, and experiments that connect intelligent systems to useful interfaces. Explore → prototype → integrate |
Model evaluation, confidence, temporal shift, and reproducible experiments. Clear questions, explicit limitations, and evidence before claims. Hypothesize → measure → iterate |
|
Shopify storefronts, payment workflows, operational dashboards, and full-stack applications built around real business needs. Design → build → ship |
Cloud services, containers, deployment pipelines, and the backend systems that keep applications running. Deploy → observe → improve |
| Recently pushed project | Language | Last push (UTC) |
|---|---|---|
| AwsamUi-jp_Makezaa Explore the repository for project details. |
HTML | 2026-10-09 |
| LPG-Distribution-Management LPG Distribution Management is a complete business management system designed for LPG distributors, godown owners, and wholesale suppliers in Bangladesh. It … |
TypeScript | 2026-10-04 |
| Typed-decision-reliability-Research-Study Explore the repository for project details. |
HTML | 2026-10-03 |
| bifrost-model-router Bifrost model router - Codex routing and model catalog management |
Go | 2026-09-24 |
| fashion-mnist-DEEP-LEARNING Explore the repository for project details. |
— | 2026-08-28 |
| usb-shield Explore the repository for project details. |
HTML | 2026-08-24 |
Updated 2026-10-10 05:50 UTC · Public, non-fork, non-archived repos, sorted by latest push. Automated pushes can also appear.
Open the curated project showcase ↗
A few public projects that connect my research interests with what I build.
| Project | What I'm exploring or building |
|---|---|
| Typed Decision Reliability ↗ ML RESEARCH |
A feasibility pilot studying confidence gates under temporal shift, using chronological splits, TF-IDF, and logistic regression. Includes evaluation, audits, and a protocol for the next stage. |
| TinyGPT ↗ DEEP LEARNING |
A small character-level transformer trained on Shakespeare, with CPU training and CLI/web generation. A hands-on exploration of language modeling. |
| Bifrost Model Router ↗ AI INFRASTRUCTURE |
A model-routing layer that connects provider catalogs and API formats, with Docker-based setup and route-aware credential handling. |
| ESP32 LLM Gateway ↗ AI × HARDWARE |
An ESP32-S3 gateway with a Cloudflare Worker, an outbound WebSocket tunnel, and an OpenAI-compatible API. |
| Newmrkt Shopify Theme ↗ COMMERCE |
A Shopify Horizon theme export for the Newmrkt storefront — part of my work on commerce experiences. |
| LPG Distribution Management ↗ BUSINESS SOFTWARE |
A distribution management application covering inventory, billing, payments, accounts, and reporting, with an AI operations assistant. |
Most used languages ↗
- Start with a question. Define the problem and what a useful result would look like.
- Make the experiment honest. Establish baselines, inspect failures, and document limitations.
- Connect research to a product. Give ideas an interface, a workflow, and a path to deployment.
- Think beyond the demo. Consider maintainability, observability, and the people using it.
I'm interested in conversations about AI/ML research, intelligent products,
Shopify, and the systems that make software useful.
Let's connect on LinkedIn ↗ · Send me an email ↗
Abdullah Al Sami / XDR-SAM / Always experimenting.