AI engineer portfolio

AI projects, case studies, and product builds

This page is a curated portfolio of AI engineering work by Muhammad Illiyin Ashraf. It covers production-minded builds across natural language processing, computer vision, generative AI, Hugging Face tooling, autonomous agents, and full-stack developer products.

Instead of listing experiments without context, each project links to a dedicated detail page with the problem, the technical approach, the implementation stack, and any live demo, GitHub repository, or related write-up that helps explain how the system works in practice.

7+ project case studiesAI agents, NLP, vision, and generative AILive demos, source links, and implementation notes

What you will find here

These projects focus on applied AI engineering rather than toy demos. The collection includes browser tools, model-serving backends, autonomous agent systems, and productized machine learning workflows built to be understandable, usable, and deployable.

  • NLP systems and text classification
  • Computer vision and multimodal tooling
  • Generative AI workflows and model UX
  • Autonomous agents, MCP, and product engineering

Related writing and services

If you want more context beyond the portfolio grid, the posts section covers implementation breakdowns and build notes, while AI services explains the kinds of freelance work I take on for startups and founders.

If you are evaluating fit, start with the individual project detail pages below. They are the strongest signal of the problems I solve and how I usually approach product, infrastructure, and ML implementation together.

Browser-based compatibility checker that detects device hardware, fetches Hugging Face model metadata, estimates VRAM and RAM needs, and tells you whether a local run looks great, tight, or too heavy.

AstroHugging FaceWebGPUHardware Detection

Open-source AI agent marketplace. Agents browse tasks, bid, submit deliverables, and earn credits. Built with Next.js, Neon PostgreSQL, and MCP.

AI AgentsNext.jsMCPPostgreSQL

Training resources for Flux.1, focusing on in-context learning capabilities and creative fine-tuning.

FluxLoRADiffusersAI

Fast Gradient Sign Method adversarial attacks on ML models. Local and cloud deployment via AWS Lambda, Amplify, and ECR with a React frontend.

PythonFastAPIAWSMLSecurity

Computer vision server that generates captions for images using AI.

Computer VisionPythonServer

Fine-tuned Llama model that rewrites weak image generation prompts into detailed, high-quality ones.

LlamaLLMPython

Interactive data structure visualizer for Bubble, Selection, Insert, Quick, and Merge sort built in C++ with sound effects.

DSAC++