B.S. Computer Science — Wayne State University
Cumulative GPA: 3.72 · Major GPA: 3.85 · Graduated: Dec 2024
Information Technology Manager · Detroit, Michigan
Hi, I'm Amer — a Bosnian-American Computer Science graduate from Wayne State University. I'm the IT Manager at Lady Jane's Haircuts for Men, keeping technology running across over 100 salon locations nationwide.
My work is real-world IT operations: LTE network infrastructure, Google Workspace, hardware and VoIP troubleshooting, websites, and the support that keeps a business running day to day. I take full ownership of the systems I manage.
Cumulative GPA: 3.72 · Major GPA: 3.85 · Graduated: Dec 2024
An iOS app that finds forgotten subscriptions by scanning Gmail for billing emails and tracks what you pay, when it renews, and what changed. Built with Expo and React Native in strict TypeScript on a Supabase backend: Postgres with row-level security, column-level grants, and a free-tier limit enforced in the database, plus Deno Edge Functions for Gmail OAuth (PKCE, encrypted refresh tokens with AES-256-GCM and key rotation), chunked and resumable inbox scans, and pg_cron-scheduled sync and push reminders. A deterministic detection engine classifies receipts, renewals, trials, and price changes, extracts amounts in 20+ currencies, infers billing cycles, dedupes merchants, and scores confidence, with no AI ever reading your email. The interface uses a custom Liquid Glass design system with spring animations, touch-scrub charts, a renewals timeline, and a WCAG-audited theme, backed by 355 unit tests, 41 database tests, and a security scan.
An executive analytics dashboard built for Lady Jane's Haircuts for Men that turns the company's year-to-date Salon Service Analytics report into an eleven-page, interactive decision tool covering 100+ salons. Built in Python with Streamlit, pandas, and Plotly, it parses and validates the raw point-of-sale export, reconciles recalculated totals against the source's own company totals, and keeps original and derived figures strictly separated. Pages include an executive command center, a salon explorer with percentile ranks and radar charts, side-by-side comparison of up to five salons, leaderboards, a US state choropleth with drill-down, service attachment and customer-mix analytics, and a year-over-year waterfall bridge. Haircut-weighted rates, a composite performance index, and an opportunity score that sizes each salon's service gap against the network median feed a rule-based insights engine. Shared filters apply across every page, and a data manager validates new uploads, flags missing or duplicate salons, and exports to CSV and Excel.
A daily sales reconciliation pipeline built for the accounting department at Lady Jane's Haircuts for Men, posting 100+ salons into QuickBooks. I designed the Excel system it runs on: a Sales Express ledger that books eight revenue and payment categories per salon and must balance to zero on every row, plus three QuickBooks upload workbooks that use dynamic-array spill formulas to turn 840 ledger lines into journal entries. A Python engine parses the raw point-of-sale export, normalizes currency formats, matches salons through an alias table, and plugs rounding so each salon balances to the cent. It drives Excel through COM automation so native formulas and filters recalculate exactly as they do by hand, then validates every file's line counts, salon alignment, dates, and reference numbers before it can be downloaded. A Streamlit front end turns the daily close into one upload and one click.
A complete rebuild of the live website for JD Capital Group, a business consulting and growth advisory firm in Birmingham, Michigan. Built from scratch in plain HTML, CSS, and JavaScript with no framework or build step: every page is a real, directly loadable document, and a custom client-side router layers app-like navigation on top. It swaps only the page content so the header never reloads, prefetches pages on hover, waits for styles and first images so nothing flashes unstyled, serves clean URLs through Vercel, and falls back to a normal page load if anything fails. The site includes case studies with photo carousels, a lightbox, and video walkthroughs, a filterable client showcase, an interactive services explorer, a serverless contact form via Web3Forms, self-hosted variable fonts, and Open Graph previews for shared links.
The COP-C Project is a project submitted to the National Security Innovation Network. It applies artificial intelligence (AI) and machine learning (ML) to battlefield communication, digitizing information that arrives through voice and text channels so it can flow into digital workflows. Our team built the Common Operating Picture Completer, an Android application in Kotlin, Java, and Python backed by a multi-agent system of machine learning models running on-device with TensorFlow Lite. Voice transcription turns spoken messages into text, then separate models handle urgency classification, sentiment analysis, and named-entity recognition (NER) to automate information extraction and turn each message into a structured military report. Data is kept in encrypted on-device RealmDB storage, alongside biometric authentication, message history, and privacy-focused display features. The result improves situational awareness and information dissemination across military units. I designed and built front-end features and UI/UX improvements, handled testing, debugging, and performance tuning, and wrote the system design and requirements documentation.
This site: a Minecraft-style title screen over a procedurally generated, rotating 3D voxel world, with no game assets used. Five biomes are generated from seeded noise in a Web Worker, meshed so only visible faces are drawn, and shaded with per-vertex ambient occlusion. Everything is rendered by a custom WebGL2 renderer I wrote in place of three.js, with procedurally painted pixel-art textures, fog, day and sunset lighting, and an instant first paint. Sounds are synthesized live with the Web Audio API, and the site has advancement toasts, saved settings, and a passcode gate. Hidden inside is a playable first-person game: walk, swim, mine and place blocks, fight zombies at night with an iron sword, and survive a full day/night cycle with hearts, fall damage, and wandering animals, on desktop or with touch controls on a phone.