its pretty good

anthropic vs open ai has reached messi vs ronaldo levels of rivalry bruv they couldn’t wait even an hour to launch lmao

transport equation ahh

crazy bro, u 2

bro i just fucking realized ure the part time farmer im connected to on linkedin HAHAHAHAH WTF

Bro wtf is going on. This post. The Jensen phone call. 50k-60k ICLR submissions. Come on!

Whoever is #malaysian, this is important, please #read and #tag anyone who may be in a position to veto this

Why I am in fact, not bullish in Malaysian Data Centers in the long term.

erdos problem #920 solved. yay. short proof & long writeup with external discoveries github.com/moseslua/erdos920

inaccurate, there aren't any ridiculous statements by malaysian netizens 🫢

scary examiner

Since I think I don't post enough on X Here are top 5 things I built/did in 2026 1. Built an end-to-end probabilistic market-microstructure framework that converts raw order-book events into calibrated, risk-aware trading decisions using forward equations, jump and diffusion models, filtering, numerical solvers, and stochastic control, using python, cpp and R. 2. Co-built a full test suite for the IMC Prosperity 4 tournament and spearheaded the development of the Options Backtest Software Implementing full Time Series Analysis and signal filtering for alpha extraction, primarily using python and cpp. (May not be preserved properly since original repo is private and I tried to transfer what I could to another with permission) github.com/moseslua/IMC-P4-P… 3. Built an MLX Inference Runtime & Quantization Pipeline, Apple Silicon LLM inference runtime implementing owned step-wise decoding, KV cache management, scheduler invariants, benchmarking, and model quantization on MLX kernels. github.com/moseslua/MLXInfer… 4. Built a generative modeling framework currently covering autoencoders, variational autoencoders, diffusion training/sampling, checkpointed training, image-directory evaluation, and reporting. GANs, score matching, SDEs, flow matching, and broader conditioning are roadmap areas rather than completed implementations. github.com/moseslua/GenLabMo… 5. Worked with a Computational Neuroscience Tech Startup and made heavy contributions to their technical software for DSP Optimzation where I centralized FFT-based band-power computation across 3 hardware bridges, replacing duplicated code with shared BandPowerComputer and rolling window buffers. Achieved 15-45x performance improvement (7.8 µs vs 117.8 µs per frame) while maintaining < 6e-14 numerical error and adding CPU/GPU/Apple backend flexibility and architected ROS 2 pipeline with multiple EEG/EMG bridges, built browser dashboard for live monitoring, electrode mapping, rule-based control, and simulation; containerized with Docker and deployed to Raspberry Pi with 60+ tests. HM: tech consulting for government law enforcement higher ups and middle level debt financial firms operating under operational distress

idk how ur aim is so bad even while using aimbot no offence @aimlab help this brother out

DoF 👁️👁️

andrew please try 茶颜悦色 cha yan yue se if you see a stall there!! legendary milk tea and i hope youll like the food in china as well, do try the braised beef dishes

bro please read their mission before anything 🫩

woah embodied ai data collection from job platforms

i am as confused as you are andrew