Skip to content
View arjunsood2025's full-sized avatar

Block or report arjunsood2025

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse

Pinned Loading

  1. llm-platform-project llm-platform-project Public

    From-scratch LLM platform: custom BPE tokenizer, transformer + Triton kernels, DDP/FSDP2 training with SFT/DPO, and a vLLM-style inference engine (paged KV cache, continuous batching, speculative d…

    Python 1

  2. quant-research-platform-project quant-research-platform-project Public

    Market-data research platform + C++ event-driven backtesting engine: partitioned Parquet/DuckDB storage, a parity-tested C++/pybind11 order-book replay engine (11.5M events/sec), and leakage-safe, …

    Python 1

  3. time-series-ml-platform-project time-series-ml-platform-project Public

    Leakage-safe training, evaluation, and serving for short-horizon prediction on noisy, regime-switching market data. Built to reduce inaccuracies when measuring signals,

    Python 1

  4. raft-kv-engine-project raft-kv-engine-project Public

    Linearizable replicated key-value store written from scratch in Rust — LSM storage engine, sans-IO Raft, and a FoundationDB-style deterministic simulator that has linearizability-checked 20,000 ran…

    Rust 2

  5. vla-robot-manipulation-project vla-robot-manipulation-project Public

    Language-conditioned vision-language-action manipulation for an SO-101 arm. BC → ACT → SmolVLA ladder benchmarked offline on held-out data, with a policy server, safety filter, and evaluation harness.

    Python

  6. isaaclab-locomotion-project isaaclab-locomotion-project Public

    PPO locomotion policies for the Unitree Go2 in NVIDIA Isaac Lab: rough terrain with a curriculum, domain randomization, and a depth-camera policy trained by teacher-student distillation. Includes a…

    Python