Akabir AbbasAI/ML and Software Engineer

Tracks sales, purchases, inventory, credit, expenses, and profit & loss while keeping all data local to one machine.

Registers models and datasets, ingests prediction logs, computes health scores, and schedules monitoring workflows with clear retraining guidance.

Combines authentication, bubble-sheet processing, barcode detection, AI-assisted essay utilities, and report generation in one operational tool.

Uses a KNN model, similar-developer matching, and tailored skill-gap guidance to make a large public dataset personally useful.

Replaces manual spreadsheet work with a focused Windows workflow for store, employee, and protection-sales reporting.

Built with the Next.js App Router and Tailwind, featuring a hand-written 2D particle engine rather than dependency-heavy WebGL.

Organizes prompts and workspaces, runs evaluation jobs in the background, and combines durable storage with fast cached reads.

Parses and groups an XML feed into useful updates, with offline-aware caching, filtering, and a sharing composer.

Pairs a Next.js interface with a FastAPI, worker, and database stack; the project explicitly frames responsible, authorized use.

Turns a topic into a complete vertical video with AI-assisted scripting, media sourcing, narration, subtitles, FFmpeg rendering, YouTube upload, retries, and channel reporting.

A Kotlin and Jetpack Compose app with Room persistence, background reminder checks, and an operational dashboard.