Cuda工作

...与准确性，同时保持训练过程的稳定性和可读性。 目前的情况 • 代码环境：Python（PyTorch） • 优化重点：提高模型准确性（训练时间和显存占用可稍后微调） • 评估方式：我已准备好一套一致的指标与验证脚本，可即时对比优化前后的表现 你需要完成的工作 1. 审阅现有实现，定位瓶颈与冗余计算 2. 重新实现或改写损失函数（含向后传播部分），确保梯度计算无误 3. 添加必要的张量操作优化（向量化、批处理、内存共享等） 4. 在我的测试集上运行并提交结果报告，其中至少包括： ‑ 指标提升幅度与对比表 ‑ 主要改动点与实现思路 ‑ 后续可扩展或进一步精简的建议 交付标准 • 指标提升需在我提供的基线之上达到统计显著 • 代码应符合 PEP-8，包含注释与简明 README • 所有修改应能在标准 GPU 环境（CUDA 11+）一次性跑通，无额外依赖冲突 如果你熟悉元学习以及高效的 PyTorch 实践，并对性能调优有系统方法，请直接告诉我你做过的相关项目、预计的优化思路与最快可投入的时间。我期待与你合作，把这段关键代码打磨到科研级水准。

... • 支持多台工业相机同步或近实时触发。 • 提供简单的设备管理界面，方便增减相机。 2. 缺陷检测核心 • 传统模块：边缘检测、纹理分析与模式识别，用于快速筛查与定位。 • 深度学习模块：可用 PyTorch/TensorFlow 训练并推理，补充复杂或模糊缺陷。 • 需要统一输出坐标、缺陷类型与置信度。 3. 系统整合 • 图像流和模型推理流程须线程安全，可连续 24 × 7 运行。 • 结果通过本地 API 或消息队列返回，以便上位机调用。 4. 交付内容 • 完整源代码与可执行文件 • 模型及训练脚本 • 简洁部署文档和使用手册 • 一份示例数据与测试报告，展示检测准确率和性能指标 只要能在实际产线上稳定运行，可自由选择 OpenCV、Halcon、CUDA 等工具。期待与你讨论数据集、照明条件及其他细节。

在特定(NVIDIA)平台上实现图像处理，应用产品是“药片判定装置”。

按我给的CUDA的属性设计线程和块 将CPU实现改编为GPU实现（如下： bool canMerge函数改写为_device_ bool canMerge void merge函数改写为_device_ void merge void findEssential函数改写为_global_ void findEssential 从CPU端初始化内存中的值如showData(dataFile,imp,minx,count); 将内存中的值传到显存 从GPU端获得 eimp[ecount] = imp[i]; eimp[ecount++].mark = false值 并传输到CPU端 画出编程的流程图

I need an Ubunt...so I can run machine-learning workloads. The job is purely the installation and configuration of everything required for the guest OS to recognise and fully utilise the GPU (nvidia-smi must work). Key points • VMware host is already installed and stable; you decide whether PCI passthrough, vGPU or another VMware feature makes most sense. • Inside the VM I will need the correct NVIDIA driver, CUDA toolkit and cuDNN ready for TensorFlow or PyTorch later on. • Once complete I should be able to launch a quick test script that confirms the GPU is visible from within Ubuntu. Deliverable A step-by-step session (screenshare or detailed command log) that leaves the VM running and the GPU operational, plus any commands I can rerun if I ever have to r...

...Video Processing * FFmpeg integration required --- ## File & Project Structure Each project should store: * Audio file * Lyrics * Scene data (JSON) * Generated images * Generated clips * Final output --- ## Advanced Features (Preferred) * Beat-synced cuts and transitions * AI-assisted prompt generation * Style consistency across scenes * Character continuity (optional) * GPU acceleration (CUDA support) --- ## Deliverables * Fully working desktop application * Clean, maintainable Python code * Installation/setup instructions * Ability to run locally without cloud dependency (preferred) --- ## Notes for Developer * This is not a simple generator — it is a **production tool** * User control over scenes is critical * Performance and stability are important * M...

...Ubuntu and I need the open-source WAN Video 2.7 stack fully installed and running on it. You will connect over AnyDesk, handle the complete setup, pull all required libraries and system dependencies, and verify the application launches cleanly with GPU acceleration enabled. During the session I will stay online to provide root access and restart the box if needed. Please be comfortable working in a CUDA-centric environment, compiling from source when binaries are not available, and troubleshooting driver or codec conflicts that sometimes appear on DGX hardware. Acceptance criteria • WAN Video Open Source 2.7 starts without errors and streams a test feed. • All supporting packages and services are documented in a short README so I can replicate the build later. &...

...Long-term work possible if successful Scope of Work: - Diagnose deep learning pipeline issues - Fix model execution errors - Debug training / inference workflow - Resolve dependency or environment conflicts - Optimize pipeline stability - Ensure end-to-end execution works correctly - Provide brief documentation of fixes Technical Stack: - Python - PyTorch / TensorFlow - HuggingFace / Transformers - CUDA / GPU acceleration - Docker / Linux environment - API integration & Data preprocessing pipeline Requirements: - Strong experience in Deep Learning production workflows - Experience debugging complex AI pipelines - Comfortable working under urgent timelines and ability to start immediately Timeline: Start: Immediately. Expected turnaround: 24–48 hours. Proposal Requi...

I’m building a camera-based ... log every reading with a timestamp, and trigger a visual or audible alert whenever negative emotions are detected repeatedly within a short window. A lightweight dashboard served with either Streamlit or Flask will let me: • watch the annotated video feed • view rolling emotion statistics and charts • review and download the timestamped log of events and alerts Optimisation for Jetson (CUDA, cuDNN, TensorRT where appropriate) is essential, and the finished app should launch from a single command, open the dashboard in a browser, sustain real-time performance, and shut down cleanly. Please keep the code modular and well commented so I can retrain or swap models later and, if convenient, provide a Dockerfile or setup script ...

Desarrollo e implementación del sistema RoadGuard (Alcance Técnico MVP según Anexo NDA). • Entorno: Desarrollo Headless (SSH/CUDA) en servidor con GPU RTX 3090. • Modelos: Implementación de YOLOv8/v10 para detección de baches, grietas y señales, incluyendo preparación de dataset y ajustes de entrenamiento. • Tracking: ByteTrack para evitar duplicidad de registros. • Georreferenciación: Sincronización de Timestamps vs Logs GPS (interpolación lineal y compensación OFFSET_MS). • Entregables: Pipeline funcional integrado y código fuente documentado.

...personalized address generator written in C/C++ and CUDA. The tool should have the following features: • Read any number of prefix-suffix patterns from the `` file (example format: `Taaa*1111` or `Tbbb*222`). • Launch a GPU kernel to continuously generate wallet addresses and compare each address with all patterns. If a match is found, write the matching address and its private key to disk. • Fully utilize GPU performance, achieving the same speed as my current test version (approximately 8 billion addresses per second). Please display a "addresses per second" counter in real-time during program execution. • Generate a plain text log file recording key events: startup time, device information, running hash rate snapshots, and each match found. ...

... "Zero-Shot" Virtual Try-On pipeline into an existing Flutter/Python e-commerce stack. Technical Stack Requirements AI/ML: Experience with IDM-VTON, Cat-VTON, or OOTDiffusion. Mastery of Stable Diffusion (ControlNet/IP-Adapter) is mandatory. Computer Vision: Expertise in MediaPipe or OpenPose (pose estimation) and DensePose (surface mapping). Backend: Python (FastAPI/PyTorch), gRPC/REST, and CUDA optimization. Frontend Integration: Flutter (Dart) for image handling and state management. Key Deliverables The "Zero-Retrain" Pipeline: A model that accepts a flat garment image and a user photo to produce a drape-accurate result without per-SKU training. Latency Optimization: Implementation of TensorRT or AITemplate to bring inference time under 3 seconds on ...

...post_content string. No Raw HTML: Mapping must use native Divi 5 module settings (Colors, Padding, Fonts, Flexbox) to ensure the layout is fully editable. Technical Stack Language: Python (FastAPI/Flask for backend, PyQt or Streamlit for local UI). Browser Automation: Playwright or Selenium (Stealth mode). OS: Windows 11. Optimization: Must be able to handle local inference calls via RTX 5090 (CUDA). Budget & Milestones ($1000 Total) Milestone 1 ($200): Functional Site Crawler (URL Listing & Selection). Milestone 2 ($400): Core Conversion Engine (Successfully importing a complex Section into Divi 5 at 100% progress). Milestone 3 ($400): Full UI Implementation, Section Slicing, and Local API Integration. Note to Freelancers: I will provide a Reference JSON file ...

...from a watch-list I will provide. Because the cameras operate 24/7 in very mixed environments—low-light corridors, exposed outdoor zones that face rain or glare, and busy high-traffic entry points—the model must remain accurate under those conditions. Solutions that leverage YOLO, TensorFlow, PyTorch, OpenCV or comparable frameworks are fine as long as they run on my existing Nvidia GPU server (CUDA-enabled). Deliverables 1. Trained model files plus any custom scripts. 2. A lightweight API or service (Python preferred) that ingests RTSP streams, performs detection, and triggers my existing alerting webhook. 3. Setup instructions and a brief validation report showing performance in the three stated conditions (night-time, outdoor weather, high traffic). I ...

... Here is what I need delivered: • High-quality masks for every image, respecting a class list that includes typical road-scene elements (road, sidewalk, vehicles, sky, vegetation, building façades, pedestrians) plus key indoor objects you would expect in a café setting (tables, chairs, walls, floor, counter). • A training pipeline in PyTorch or TensorFlow that I can run on Ubuntu 22.04 with CUDA, along with a clear README covering dataset preparation, training, and inference. • A model that reaches at least 0.75 mIoU on a private test split I will share once the annotations are complete. You are free to use tools such as CVAT, LabelMe, Detectron2, DeepLabV3+, SegFormer—or any comparable framework—as long as the final workflow remain...

...into the transcript with millisecond accuracy. Both real-time feedback (small overlay suggestions) and post-video analytics (downloadable PDF/CSV plus on-screen dashboard) are needed. I’m happy for you to build with tools such as OpenCV, MediaPipe, TensorFlow, PyTorch, spaCy or similar—use what you are fastest with as long as the models run efficiently in a web environment (GPU acceleration via CUDA or WebGL is a plus). Deliverables 1. Source-controlled codebase ready to deploy on a standard cloud stack (Docker image or Heroku-style procfile). 2. Front-end UI (React, Vue or vanilla JS) that lets users toggle between real-time and upload modes. 3. Modular inference services for vision and audio that can be retrained or swapped if I add new metrics later. 4. C...

...data and route only the most promising parameter sets back to the gate model. Latencies must stay sub-millisecond from signal to order, so a coherent design for GPU–FPGA–QPU orchestration is essential. Deliverables • A documented architecture diagram showing data flow between classical AI, middleware, and the chosen quantum SDK (Qiskit, Braket or similar). • Clean, modular Python code with C++/CUDA kernels where latency demands it, fully containerised for reproducibility. • Back-test and forward-test reports on at least one major FX pair and a US equity futures contract, including Sharpe, max drawdown, and execution slippage statistics. • Deployment guide for a colocation environment, covering queue management to the quantum back-end and f...

请注册或者登录后来查看详情。

...job is to create the complete vision-detection module—from model training or fine-tuning through to a clean ROS 2 node that subscribes to an image topic and spits out the detected objects with bounding boxes (or masks) and a confidence score. OpenCV, TensorFlow/PyTorch and any of the common ROS 2 image-transport plugins are all fine as long as the final node runs on Humble and stays GPU-agnostic (CUDA acceleration is a bonus, not a requirement). I already have a test rig with a standard USB camera; if you need specific calibration images I can capture them for you. Please deliver: • Source code for the detection model and ROS 2 node • A launch file that brings everything up with default parameters • A brief README explaining setup, parameters and expect...

...environment that emulates Jet Nano hardware for research and development on machine-learning models. The goal is to give my team a sandbox where we can move seamlessly from data preprocessing and feature extraction through model training, evaluation, deployment, and monitoring—without touching the physical board until we are ready. Here’s what I need: • A reproducible simulation that mirrors Jet Nano’s CUDA-enabled GPU, memory constraints, and I/O. • Containerised tool-chain (PyTorch, TensorRT, cuDNN, etc.) with scripts that cover the full life-cycle: preprocessing, training, hyper-parameter sweeps, evaluation metrics, and a mock-deployment stage that tracks resource usage and latency. • Clear documentation so any teammate can spin up the en...

... • Accepts at least JPEG files for input; adding PNG or BMP later should remain possible. • Generates a short video (MP4 preferred) by feeding the image through Stable Diffusion and WAN2.6. • Interface must feel intuitive for non-technical users while exposing advanced settings in an “expert” panel. • Conversion speed is critical; please optimise GPU utilisation and let me choose device (CUDA / DirectML). • Output parameters—resolution, frame rate, length, prompt text, CFG scale, seed—should all be editable before rendering. Deliverables 1. Executable installer (or portable folder) with all weights and dependencies bundled for offline use. 2. Source code with clear build instructions so I can re-compile if models up...

Backend for a my app using FastAPI, WebSock...and concurrency: Comfortable designing and debugging async workflows. Hands‑on AI integration experience with at least one of: Whisper STT or other speech‑to‑text engines. LLaMA/transformer‑based LLMs or OpenAI‑style APIs. TTS systems such as Coqui, Kokoro, or Piper. Realtime systems: WebSockets, WebRTC, or other low‑latency streaming architectures. Nice to Have GPU & deployment experience: CUDA, GPU environments, and performance tuning (CPU vs GPU). Docker, nginx, PM2, and production deployment pipelines. Background processing: Job queues/workers for heavy audio/video processing. Experience orchestrating long‑running media/AI tasks. Video processing tools: FFmpeg, Wav2Lip, or similar for video generation and post‑processing.

...Predict response to therapy (Responder / Non-responder) Predict survival category Predict recurrence risk For MVP: Start with diagnosis, then add treatment prediction. STEP 2: Setup Development Environment Install Dependencies Python 3.9+ PyTorch MONAI pydicom numpy scikit-learn FastAPI or Flask Example: pip install monai torch torchvision pydicom fastapi uvicorn scikit-learn Setup GPU Local CUDA GPU OR Cloud (AWS/GCP/Azure) STEP 3: PET Scan Dataset Preparation Collect Dataset Public PET database (e.g., TCIA) Research partnership dataset Must include: PET images Diagnosis labels (Optional) treatment outcome labels Organize Data Structure: data/ train/ val/ test/ Handle DICOM Files Use pydicom to read images Convert to 3D tensors Normalize voxel intensity STEP 4:...

...training worker (Docker, from scratch) - PHP/MySQL licensing backend + Stripe webhook integration - Unified cross-platform installer (detects DAWs, installs everything in one pass) - GitHub Actions CI/CD (Windows + macOS builds) - Full Apple + Windows code signing pipeline - Documentation (User Guide + Developer Guide + BYOK Setup) Key technical requirements: - CPU default with automatic NVIDIA CUDA detection for Live Mode - RMVPE primary pitch extraction + user toggle (Harvest/Crepe/FCPE) - High-quality resampling (44.1k-96k) in C++ wrapper - AI Cleaning (de-reverb/isolation) in front of inference chain - Index Rate + .index file exposed in UI/API - Batch processing via ZMQ socket bridge Terms agreed: - Budget: $2,500 (6 milestones) - Timeline: 6 weeks (Feb 23 - Apr 3, 2026) -...

...similar) - Weasyprint or ReportLab for PDF - Typer CLI with subcommands: - transcribe - diarize - lesson-report - aggregate - YAML config file - Logging, progress bars, caching (skip if output exists), error handling Deliverables: - Full repo structure - All source code (src/ layout, CLI, config, prompts, PDF renderer) - Installation instructions for Windows 11 (Python, ffmpeg, Poetry, CUDA) - Example commands - Test guide with sample audio Please show experience with WhisperX / faster-whisper, Pyannote, Ollama, and Weasyprint on Windows + GPU setups in your proposal. Thank you! Vladimir...

...rapid target motion • Adapt to scale and orientation changes • Maintain lock under partial occlusion • Recover gracefully if tracking confidence drops • Avoid drift over time A re-detection or hybrid tracking strategy is preferred if it improves robustness. Technical Requirements Preferred stack: • Python + OpenCV OR C++ + OpenCV • Modular architecture • Hardware acceleration support (CUDA / TensorRT) is a strong plus • Experience with: • Siamese-based trackers • DeepSORT-like approaches • Hybrid detection + tracking pipelines Clean, well-documented code is mandatory. Deliverables 1. Fully functional Linux application 2. Source code repository 3. Setup instructions + dependency list 4. Short demo video...

...website. I have the hardware available but need an expert who can install the model, configure all dependencies, and expose an endpoint that my front-end widget can call. Here is what I have in mind: • Select and download an open-weight GPT-like model that can reasonably run on local hardware (e.g., Llama-2, Mistral, or another suitable alternative). • Set up the execution environment—Python, CUDA, PyTorch or TensorFlow—plus any supporting libraries (LangChain, FastAPI, uvicorn, etc.). • Create or refine an inference script that keeps response times low enough for smooth chat. • Build a lightweight API (REST or WebSocket) so the website can pass the user’s prompt and receive the model’s reply. • Hand me clear, repeatable...

...expectations • Real-time theft detection logic that raises an event or REST webhook the moment a suspicious removal is spotted • On-screen bounding boxes and confidence scores for detected grocery items and customers • Continuous customer counter with hourly CSV/JSON export • Installers or scripts for Windows 10/11 and Raspberry Pi OS, including all required Python, OpenCV, PyTorch/ONNX, CUDA (where available) dependencies • A simple dashboard that shows live feed thumbnails, current customer count, and the last N theft alerts • Clear instructions on adding new grocery SKUs later Acceptance will be based on: 1. Smooth 25-30 fps inference on 1080p streams under Windows with GPU, and ≥10 fps on Raspberry Pi using CPU or a USB accele...

...the hardware allocated and wallets ready; what I need is an engineer who can take the nodes from zero to profitable operation and then keep them humming. Key tasks • Provision and secure each H100 instance, configure networking, firewalls, SSH keys and wallets • Containerise the stacks with Docker (Kubernetes or Podman are possible later, but Docker is fine for the first iteration) • Tune CUDA-level settings so every GPU cycle counts and rewards are maximised • Build simple Bash or Python scripts that monitor logs, restart on failure and push basic alerts • Produce step-by-step documentation so the setup can be replicated or audited at any time Acceptance criteria • Nodes reach consensus, stay above 99 % uptime and begin generating rewa...

Lead AI / Fullstack Engineer — ...communication. ​Traffic Localization: Optimize routing protocols to maximize performance within the TAS-IX network. ​Candidate Requirements ​AI / ML Engineering: ​Proven experience with End-to-end (E2E) speech models (Moshi, AudioLM, or similar). ​Deep proficiency in PyTorch and Transformer architectures. ​Hands-on experience in Fine-tuning LLMs/S2S models for new language groups. ​Expertise in CUDA 12.x and NVIDIA optimization libraries. ​Fullstack Development: ​Expert-level knowledge of WebRTC / WebSockets for real-time media streaming. ​Demonstrated experience in developing Telegram Mini Apps (TMA). ​Professional mastery of FastAPI and React / Next.js. ​Strong understanding of the constraints and requirements of Low-latency systems.

This project requires real GPU computation, correct Bitcoin cryptography handling, and verifiable results. This is not a demo or theoretical project. The program must be fully functional and tested. Only apply if you have proven experience with CUDA, cryptography, or Bitcoin key handling.

...modern GPUs and expose a clean, future-proof API for downstream applications. My end goal is to abstract away vendor-specific quirks so a data-scientist, graphics engineer, or researcher can tap into raw parallel power without worrying about whether the machine is running Windows, Linux, or macOS, or whether it ships with NVIDIA, AMD, or Intel silicon. You’re free to recommend the optimal blend of CUDA, ROCm, OpenCL, Vulkan, or even a custom compute layer—what matters is performance, portability, and clean code that’s easy to extend. I’m open to focusing on a single workload first (machine-learning kernels, real-time graphics effects, or heavy scientific simulations) if that helps us validate the core, then scaling outward. Deliverables I’m exp...

Job Title: CUDA Developer Needed – GPU-Accelerated Bitcoin WIF Key Recovery Tool (Verification Required) Project Description: I am looking for an experienced CUDA / GPU developer to build and optimize a high-performance Bitcoin WIF private key recovery program. This project requires real GPU computation, correct Bitcoin cryptography handling, and verifiable results. This is not a demo or theoretical project. The program must be fully functional and tested. Only apply if you have proven experience with CUDA, cryptography, or Bitcoin key handling. Technical Requirements: - Written in C++ with CUDA - Runs on NVIDIA GPUs - Command-line interface (CLI) - Supports Bitcoin WIF (Base58Check) - Supports compressed and uncompressed private keys - Correct che...

Lead AI / Fullstack Engineer — ...communication. ​Traffic Localization: Optimize routing protocols to maximize performance within the TAS-IX network. ​Candidate Requirements ​AI / ML Engineering: ​Proven experience with End-to-end (E2E) speech models (Moshi, AudioLM, or similar). ​Deep proficiency in PyTorch and Transformer architectures. ​Hands-on experience in Fine-tuning LLMs/S2S models for new language groups. ​Expertise in CUDA 12.x and NVIDIA optimization libraries. ​Fullstack Development: ​Expert-level knowledge of WebRTC / WebSockets for real-time media streaming. ​Demonstrated experience in developing Telegram Mini Apps (TMA). ​Professional mastery of FastAPI and React / Next.js. ​Strong understanding of the constraints and requirements of Low-latency systems.

manual intervention. 3. Re-assemble processed frames back into a single clip using FFmpeg (or similar), ensuring temporal consistency—no flicker or dropped frames. 4. Expose a simple CLI command such as: python --input --output --strength 0.7 --seed 42 5. Provide a short README covering environment setup (Python, diffusers / transformers versions, CUDA requirements), example usage, and expected runtimes Acceptance criteria • The script completes a sample without errors and produces visibly live-action styling throughout. • Code is clean, commented, and includes a or environment.yml. Delivery: source code, README, and one converted sample clip produced by your wrapper.

...short written walkthrough covering hardware requirements, model parameters, and tips for further tuning. Acceptance criteria 1. Frame-by-frame identity preservation ≥ 95 % (verified with face-recognition scores). 2. No temporal flicker visible on 30-fps playback. 3. End-to-end generation time under 2× video length on a single high-end GPU. Tech stack keywords: PyTorch, TensorFlow, FFmpeg, CUDA, Google Colab, facial-landmark detection, GAN inversion. Roadmap beyond this delivery Once the core system is proven, I plan to expand into other AI-driven video features—scene synthesis, automated dubbing, even real-time object tracking—so clean, well-documented code is essential for future extension. Ready to start as soon as we agree on the approach, and ...

...- Auto-sync narration and visuals - Options: - voice selection (male/female) - narration speed - background music (optional) - subtitles (optional) Tech Requirements: - Must support OFFLINE mode (local machine) using open-source models (preferred) - Should also support ONLINE mode (server/cloud deployment) - Efficient pipeline (render without crashing) - Works on CPU + GPU if available (CUDA GPU preferred) Preferred Implementation (engineer decides exact tools): - Python backend (FastAPI preferred) - Local model inference pipeline - Video assembly using FFmpeg / MoviePy - Open-source TTS narration (example: XTTS, Piper, Coqui TTS, etc.) - Open-source image generation or whiteboard assets pipeline - LLM for storyboard/script breakdown (open-source model OR cheapest API) ...

...corre sin interrupciones perceptibles. El proyecto consiste en el desarrollo de un motor de inferencia de alta performance para la detección, clasificación y seguimiento de múltiples clases de objetos en entornos dinámicos complejos, utilizando hardware dedicado. Implementación de arquitecturas de detección (YOLO/RT-DETR) y algoritmos de tracking.• Optimización de modelos para hardware NVIDIA (CUDA/TensorRT).• Fusión de datos provenientes de sensores externos (Telemetría/GPS) con flujos de video 4K.• Desarrollo de lógica de persistencia en bases de datos geoespaciales. • Seniority comprobable en Python y OpenCV. Experiencia en el ciclo completo de vida de modelos de visión: desde ...

...advise on the best hardware stack to achieve the 15-50 person real-time requirement: Cameras: Recommend specific sensors or cameras (e.g., Wide FOV, Global Shutter, or IR-capable for low-light/glare robustness). Processing Units: Advice on edge deployment. Can this run on a Raspberry Pi 5 with an AI Kit, or is an NVIDIA Jetson (Orin/Nano) required? If commodity GPUs are needed, specify minimum VRAM/Cuda core requirements. Kits: Recommend specific "plug-and-play" kits or enclosures suitable for the indoor/outdoor environment described. Final Deliverables Hardware Recommendation Report: Detailed list of suggested cameras, lenses, and processing kits (Raspberry Pi, Jetson, etc.) tailored to this specific use case. Source Code: Ready to plug into a Python environment ...

I’m running ...trade-offs you introduce so I can reproduce and benchmark I already run basic line-profiler and torch-autograd checks, so I’m looking for deeper insights—vectorised ops, smarter batching, async data movement, or architectural tweaks I may have missed. Feel free to use tools like PyTorch Profiler, nvprof, or your preferred optimisers as long as the final instructions remain reproducible in a standard CUDA environment. If that sounds straightforward, let me know your availability and how you’d approach the first pass; I’m ready to share the repo right away.

..., Twilio or Meta API) OR a custom Mobile App (Flutter/React Native) for security staff. Dashboard: A simple web-based or local interface to view live logs, replay detected incidents, and manage sensitivity settings. Technical Requirements: Programming Language: Python. Frameworks: PyTorch, TensorFlow, OpenCV, YOLO (v8/v10), or MediaPipe. Hardware Compatibility: Must be optimized for NVIDIA CUDA cores / TensorRT. Scalability: The code should support multiple camera streams simultaneously. Deliverables: Full Source Code (well-documented). Setup Guide (How to install on the NVIDIA device and connect cameras). A working prototype/MVP demonstrating the detection of basic theft actions. Ideal Candidate: Proven experience in Computer Vision and Action Recognition. Previous ...

I need a Windows-based GPU workstation dedicated to running local large-language-model workflows. I need someone who can walk me through the full setup—hardware , CUDA drivers, PyTorch/TensorFlow installs, plus the extra tools I rely on for text-to-video generation and similar AI workloads. Your first task is to get the machine fully operational: verify BIOS and power settings, install the latest GPU driver stack, configure CUDA/cuDNN, and deploy the core frameworks. From there we’ll layer in local-LLM utilities (e.g., , Ollama) alongside Stable Diffusion or any other video-generation packages I might explore. Clear, repeatable documentation of every step is essential so I can reproduce the environment later. Once the base system is stable, I’d like on...

...3), OCR (paddleocr 2.10.0 on paddlepaddle 3.0.0 / paddlepaddle-gpu 2.6.2), and post-processing with scikit-learn 1.6.0. Although one GPU-ready wheel is present, all processing still executes on the CPU. The goal is full NVIDIA CUDA utilisation across the entire workflow, from frame decoding to final inference. I need you to: • Profile the current code, pinpoint CPU-bound sections, and migrate or rewrite them for GPU execution (CUDA, CuDNN, cuBLAS, or other relevant CUDA-based APIs). • Update or swap libraries where necessary—feel free to recommend faster CUDA-compatible alternatives if they will not break accuracy (e.g., CuPy, TensorRT, NVIDIA Video Codec SDK). • Modify the code so GUI-less batch processing and real-time video run...

...spots a potential anomaly. All processing must happen in real time without introducing perceptible latency to the surgeon’s view. My current hardware outputs standard HDMI and records to DICOM, so your code should sit either between the camera head and the display (FPGA, GPU box, or high-performance PC is fine) or run as a software module on the workstation already attached to the scope. OpenCV, CUDA, TensorFlow, or similarly robust libraries are welcome—just keep licensing constraints clear. Deliverables • Executable or deployable source that enhances image clarity, performs real-time analysis, and triggers automated anomaly detection. • API or integration hooks so I can feed the processed stream back to my recording software. • A concise user gu...

...Engineer - Real-time Edge AI (OpenCV, ONNX, CUDA & TensorRT) Busco un Senior Computer Vision Engineer con experiencia demostrable en Edge AI para desarrollar un sistema de asistencia táctica en tiempo real basado en captura de vídeo externa. Desafío Técnico Principal: El sistema debe procesar un flujo de vídeo HDMI, realizar OCR de alta precisión y detección de objetos, y consultar una base de datos local con una latencia end-to-end inferior a 100ms. Stack Tecnológico Requerido: • Lenguaje: Python 3.10+ con arquitectura OOP escalable (Interfaces abstractas). • Visión: OpenCV avanzado y procesamiento de imágenes para OCR de stacks y botes. • Motores de Inferencia: Experiencia obligatoria con ...

...machine freeze during model training? Welcome to your new digital superpower. I bridge the gap between your ideas and the raw power of Microsoft Azure. I don’t just "rent servers"—I architect secure, high-performance environments so you can focus on building the future. What’s in my secret sauce? GPU Beasts: Access NVIDIA N-Series (V100, A10, T4) for AI/ML. Ready-to-Go Stack: I’ll pre-install CUDA, PyTorch, TensorFlow, or Docker. No more driver headaches! Fort Knox Security: Advanced Firewalls & private VPNs. Your VM stays invisible to the public web. Windows or Linux? I speak both. Whether you need an RDP or an SSH terminal, I’ve got you. The "Date Before You Marry" Trial Not sure if the speed is right? For just $5, ...

...with a brand-new RTX 5090 and need TensorRT installed, tuned and ready to accelerate Stream Diffusion inside TouchDesigner. I haven’t settled on a specific release yet, so I’ll rely on your guidance to pick the most stable, future-proof version (including matching CUDA and cuDNN builds) for this GPU. Here’s what I expect: • Recommend the best TensorRT version for an RTX 5090 Windows environment and explain why it’s the right fit. • Handle the full installation: download packages, configure environment variables, and verify driver / CUDA compatibility. • Prove the install works by running a sample inference, then confirm TouchDesigner can see the TensorRT engine for Stream Diffusion. • Leave me with a concise, step-by-step recap ...

...a brand-new RTX 5090 and need TensorRT installed, tuned and ready to accelerate Stream Diffusion inside TouchDesigner. I haven’t settled on a specific release yet, so I’ll rely on your guidance to pick the most stable, future-proof version (including matching CUDA and cuDNN builds) for this GPU. Here’s what I expect: • Recommend the best TensorRT version for an RTX 5090 Windows environment and explain why it’s the right fit. • Handle the full installation: download packages, configure environment variables, and verify driver / CUDA compatibility. • Prove the install works by running a sample inference, then confirm TouchDesigner can see the TensorRT engine for Stream Diffusion. • Leave me with a concise, step-by-step rec...

Looking for developer who can work on below requirment . Lead design and implementation of GPU computers for deep learning; optimize algorithms and mentor team Must have key skills cuda,c++,Gpu Programming Other key skills Parallel Computing,Opengl,Opencl Job description What you’ll do CUDA is a must JD For Senior / Lead Engineer (HPC GPU):- As a Senior / Team Lead (HPC) you will provide leadership in designing and implementing groundbreaking GPU computers that run demanding deep learning, high-performance computing, and computationally intensive workloads. We seek an expert to identify architectural changes and/or completely new approaches for accelerating our deep learning models. As an expert, you will help us with the strategic challenges we encounter, includi...

...virtually no perceptible delay. The tool must lock on to faces accurately, track expressions, match lighting and color, and render the composite at a stable frame rate suitable for streaming or studio recording. The core pipeline should include high-resolution face detection, landmark tracking, real-time inference with a modern GAN or transformer model, and seamless blending. Feel free to lean on CUDA-accelerated TensorFlow or PyTorch, OpenCV for image I/O, and any efficient post-processing libraries you trust—what matters is rock-solid performance and visual fidelity. I want the interface to be simple: a preview window, a slot to load or capture the target face, quick toggles to enable/disable tracking, and an option to record or pipe the output to a virtual camera devic...