Cloud & Mobile Engineer — Carnegie Mellon University, Africa
Building production-grade mobile applications and scalable cloud infrastructure. From Flutter apps live on the Play Store to native Windows desktop tools shipped on the Microsoft Store — I ship things that work.
Career
Developing the mobile app codebase and features using clean, scalable, and maintainable code for smart assistive devices (crutches, prosthetics, wheelchairs)
Integrating and maintaining Bluetooth/BLE connectivity between the mobile app and smart assistive hardware
Implementing real-time sensor data display — weight sensors, posture, movement, and activity insights for patients and clinicians
Building patient and clinician dashboards with secure, stable data synchronization between device, app, and Firebase/Supabase backend
Maintaining modular architecture and GitHub version control to support future smart assistive device integrations
Thorough testing on Android and iOS, bug resolution, and performance improvements across the full app lifecycle
Configured network infrastructure — rack setup, wiring, and router connections for enterprise deployments
Resolved complex LAN connectivity issues and monitored network reliability across multiple sites
Customized WordPress themes & plugins for high-profile government websites and portals
Performed security audits and cross-browser functionality testing across all managed platforms
Work
Shipped Software
Six native Windows applications published under Cresco Systems Labs, live on the Microsoft Store. Each one talks directly to the machine it runs on — reading battery and thermal sensors at the hardware level, driving GPU-accelerated codecs, gating filesystem actions behind human approval, or processing documents fully on-device — rather than treating Windows as a black box.
An AI prompt engineering toolkit for Windows that helps you craft, test, and refine prompts for large language models (LLMs) like GPT-4, Claude, and Gemini.
Local AI Playground. Experiment with different LLMs and parameters directly within the app, ensuring privacy and rapid iteration without constant API calls.
Powerful prompt editor with syntax highlighting and version control
Organize prompts into collections and easily export/import them
Supports popular LLMs: GPT-4, Claude, Gemini, and more
A system-tray-resident power optimizer for Windows 10/11 that gives always-on visibility into every watt a laptop or desktop consumes, with one-click tools to claw back battery life.
Reads hardware directly. Battery voltage, charge current, and per-core CPU temperature are pulled straight from the sensor layer (with a WMI fallback) — not estimated from OS averages — and refreshed every 2 seconds.
Per-process wattage profiler with one-click suspend / throttle of background drains
Green Mode — six toggleable optimization layers (display, CPU governor, GPU rail, NVMe APST) for up to +3h 20m runtime
Built on WinUI 3 + Windows App SDK with Mica material and live tray tooltips
An all-in-one media tool for Windows that downloads video and audio from thousands of sources, then converts and organizes everything offline from a single, clean interface.
GPU-accelerated transcoding. Format conversion is offloaded to NVIDIA NVENC and Intel Quick Sync encoders on the host GPU, dramatically cutting conversion time versus CPU-only software encoding.
144p–4K downloads, full playlist/channel grabs, resumable queued downloads
Batch format conversion across MP4 / MKV / WEBM / MP3 / FLAC with bitrate & codec control
Built-in local media library with search, preview player, and one-tap re-conversion
Converts PDFs, Word, PowerPoint, Excel, HTML, and images into clean, structured Markdown — built for developers feeding documents into AI pipelines, RAG systems, and vector databases.
100% on-device processing. Every conversion runs locally on the machine's CPU via Microsoft's open-source MarkItDown library — no file ever touches a network socket or cloud server.
Strips layout noise (headers, footers, page numbers) that inflates RAG token counts
Batch-converts entire folders in one pass, preserving tables, lists, and code blocks
Zero telemetry, zero uploads — fully offline by design
A Windows desktop agent that turns natural-language prompts into verified local actions — file moves, renames, and terminal commands — routed across multiple AI models with a local Ollama fallback for fully offline runs.
Security-gated by design. Zero terminal commands execute and zero files move without an explicit checkbox approval — every action ships with one-click undo and instant rollback.
Multi-model AI routing with a local Ollama fallback when offline or API-limited
Human-in-the-loop approval gate before any command runs or file is touched
Guaranteed one-click rollback to instantly revert any directory modification
A native Windows CRM built for marketers and sales professionals who sell on relationships, not spreadsheets — replacing sticky notes and forgotten follow-ups with a single pipeline that tells you exactly who to talk to today, and what to say.
Private by design. Your pipeline lives on your device — no forced cloud account, no data mining, and no subscription required to keep your own contacts.
Every contact auto-ranked Hot, Warm, or Cold with a built-in follow-up cadence — 2–4 days, 7–14 days, or 30–90 days
Signature coaching engine enforces a 3:1 value-to-ask ratio, flagging asks made before enough value has been given
Value-drop message templates, one-click CSV reporting, and 18+ language / currency support for global sales floors
Competencies
Background
Get In Touch
Available for roles, freelance projects, and conversations about mobile, cloud, and what's possible with modern engineering.