Skip to content
View jatin711-debug's full-sized avatar
🤪
I may be slow to respond.
🤪
I may be slow to respond.
37 followers · 75 following

Achievements

Achievement: QuickdrawAchievement: YOLOAchievement: Pull SharkAchievement: Public Sponsor

Achievements

Achievement: QuickdrawAchievement: YOLOAchievement: Pull SharkAchievement: Public Sponsor

Block or report jatin711-debug

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

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

Report abuse
jatin711-debug/README.md

Building high-performance storage engines, distributed event systems, and GPU-accelerated ML infrastructure from first principles.

⚡ Featured Systems & Engineering

🦀 TurboSuperMemory

High-performance cognitive memory & vector storage engine written in Rust & CUDA.

  • 32× TurboQuant Compression: 1M vectors in ~61 MB using Polar Fast Walsh-Hadamard Transforms (FWHT) + Lloyd-Max codebooks.
  • 2-Stage Hybrid Retrieval: Sub-ms SIMD candidate scan + Stage-2 token MaxSim late interaction (ColBERT).
  • Cognitive Graph: In-memory spreading activation, ACT-R power-law decay, and NLI-based belief revision with $0.00 write-time LLM cost.

Rust CUDA C++ SIMD PyO3 gRPC / Axum

🗄️ CronosDB

Timestamp-triggered pub/sub database engineered for precise time-scheduled event delivery.

  • At-least-once delivery semantics with persistent event replay.
  • Custom file-backed storage engine with low operational overhead.
  • High-throughput gRPC subscriber pipeline & scheduling engine.

Go gRPC Kafka PostgreSQL SQLC

🩺 Medical Multimodal AI

Vision-language fine-tuning & explainable medical imaging pipelines.

  • Fine-tuned Med-GEMMA models for pneumothorax and clinical diagnosis.
  • Attention heatmap extraction and GPU memory footprint optimization.

PyTorch Med-GEMMA HuggingFace CUDA

🎥 GPU Video AI Pipelines

Low-latency hardware-accelerated media preprocessing and inference.

  • Hardware-accelerated decoding/encoding via NVIDIA NVENC/NVDEC.
  • High-throughput frame extraction for vision-language models.

C++ Python FFmpeg OpenCV NVIDIA Video Codec SDK

🧰 Core Toolchain & Technologies

• Systems & Storage : Rust, Go, C++, CUDA, SIMD/AVX2, PostgreSQL, Kafka, Redis, SQLite/redb
• ML & Inference    : PyTorch, vLLM, Triton Inference Server, SGLang, SentenceTransformers, HuggingFace
• Infra & Media     : Docker, Kubernetes, Linux, gRPC, Axum, FFmpeg, NVIDIA NVDEC

Pinned Loading

  1. cronos_db_golang cronos_db_golang Public

    Distributed Timestamp-Triggered Database with Built-in Scheduler & Pub/Sub

    Go 10

  2. TurboSupermemory TurboSupermemory Public

    Cognitive Memory Engine for AI Agents

    Rust

  3. Project-ARGUS Project-ARGUS Public

    Edge-compute AI pipeline that runs in orbit.

    Python 3 1

  4. TritonServerGRPCTesting TritonServerGRPCTesting Public

    Triton YOLO11x Inference Server & Client

    Python

  5. PocketTitan PocketTitan Public

    External-memory post-training quantization engine for extreme-scale LLMs & MoEs (DeepSeek, GLM) on consumer GPUs under 3.5GB VRAM.

    Python