Benjamin R.

In january 2026 my opensource Bellsack game engine briefly went viral. My fans and market speculators created 2 memecoins of the same name which moved over 2 million dollars in 1 day and at one point had $700k, but soon crashed to near $0, not my money but I got $13k in transaction fees. I am hoping to do it bigger next time as I continue improving my product. I am available to do a variety of programming, math, or other work. My upwork success score seems to be low recently because of lack of doing upwork jobs. I will get your work done. I am available for general programming or math jobs. For example... I can make a web browser do almost anything, but I dont know alot of libraries. I've used up to 80 off-screen canvases at once and over a trillion multiplies per second (teraflop) in browser, neuralnets, webcam, gamepad, ajax, low lag connection to servers, generating files that didnt even exist for download from a lone html file, opening and creating zip files, bending pics onto curved surfaces, 2d voxels, time rewind/forward, physics sims, scalar fields up to 15000 dimensions, divs (dom nodes) at angles moving around like with springs between them, putting webpage parts onto 3d polygons, custom designed programming languages built from scratch, secure-hashing and encryption, universal-pattern-calculus-combinators, drag and drop, searchable tables sortable by columns or by custom tuneable orders by sliders and checkboxes, data structures for DAG based secure sync across untrusted borders, optimization, connection to other systems. Just about anything you want a browser to do, that is defined purely in terms of computer behaviors (not specific other systems since theres alot of them and I likely dont know all of them you need, though can learn), I can find a way to do it. Education: MS 3.23, BS 3.55, Computer Science, NC State Univ 2000-2006 I can do up to 25 hours/week on average. CAN USE AI-GENERATED CODE OR NOT ON YOUR PROJECT: On some projects I use small pieces of AI generated code but I do read and understand and improve the code before using it in a final project. Is that ok, and to what extent here? Its very good with math. If that is not ok I will just learn things in general from AI but not use any of its code output. The generated code is copyrighted to me by the OpenAI license "We hereby assign to you all our right, title, and interest, if any, in and to Output." so I can legally sell or opensource that code, if its not just a big quote of something else the AI saw. Its very smart and comes up with original ideas. For example, my "computeUint10Distances = (tridist, shift, yFrom, yTo, xFrom, xTo, orthogonalDist, diagonalDist, optionalMaxDist)" function in Blob Monsters Game was written part by me and part by AI. Experience - Highlights: • (in progress) My own programming language. All possible programs are made of combos of calls of a single function for low-lag GPU-speed secure Internet sync. Tiny or huge pieces of computing securely realtime copy/pastable across internet. • (in progress) Browser webcam and GPU.js will turn tennis balls into 6d joysticks Billions of people have no VR-like controllers. A bucket of tennis balls is cheap. .js file that any webpage can use. Desktop or hold balls above phone front camera. • Variety of languages and systems across jobs and my open-source projects. • Talking advanced things with nontechnical people. • Experimental search engine made of 100k wikipedia images and 70gB of text. • Reverse-engineered SSL code to make a program compatible with packet sniffers. • Stateless form of OpenCL does depend-net of GPU work in 1 low-lag function • Used 400 VM computers at once • Phone takes picture of bill (power, phone, etc), automaticly scans into bank. • Volume norming where holding microphone closer to speaker than your mouth cant explode while amplifying a whisper from across the room. • Musical instrument evolver generates java code that computes wave amplitudes • My JSON parser/generator is very small: 16kB • Deep debugging in Eclipse – once traced a network call for 2 hours • Reduced lag by doing many SQL queries in the same network call • Textarea to change learning algorithm code while its learning and instantly see • Language detection of top 35 languages by curved ngram statistics. • Reordered complex caching steps to extend map search to variable distance • Tool to automatically verify a unicode upgrade • Computed top 16 Wikipedia centroids contains Albert_Einstein and Molecule • Packaging java programs and any files into an executable jar file. • Custom webcrawlers (1) find many files types (2) cache restaurant menu internals