Pytorch jobs
...automated-bidding module so buyers can set ceilings and let the system compete for them. • A fraud-detection layer that flags suspicious listings or bidding patterns before a transaction closes. Everything must sit behind a clean, minimal interface—no clutter—because both food and pharmaceutical professionals will be using it every day under time pressure. I’m open on stack, but Python with TensorFlow/PyTorch for the models and a lightweight React or Vue front end makes sense; CUDA optimisation is a plus for the Inception pitch. Acceptance criteria for the MVP: 1. End-to-end workflow: seller upload → AI price suggestion → live auction → automated & manual bids → secure checkout. 2. Latency for price predictions under two seconds...
...visual features and assigning each image to the correct category. My end goal is straightforward: a reliable model that can classify every new photo I add, with clear metrics that show its accuracy. I haven’t fixed the exact number of categories yet; we can review the sample set together and decide what makes sense. You are free to work in Python with libraries such as TensorFlow, Keras or PyTorch, and any supportive tools for preprocessing, augmentation, and evaluation. I’ll share the image folders via a cloud link once we start. Please hand over: • Clean, well-commented source code (notebook or script) • The trained model weights or a reproducible training script • A short report summarising accuracy, confusion matrix, and key insights I...
...want to build a small AI-driven agent that concentrates purely on weather predictions, with rainfall and precipitation as its only inputs, so I can decide exactly when to harvest. Here is what I need the system to do for me: • Pull historical and real-time rainfall data from reliable public APIs (NOAA, Meteostat, or a source you recommend). • Train a lightweight model in Python—TensorFlow, PyTorch, or an equally proven library—that forecasts field-level precipitation a few days ahead. • Convert the raw forecast into a simple recommendation: safe-to-harvest, harvest-soon, or delay. • Present that recommendation in a clear format. A small web dashboard is ideal, but a command-line script with a clean JSON output is acceptable if it speeds ...
...configurable hot-keys and an on-screen overlay for feedback—but I need an experienced developer to design and train the hand-gesture recognition pipeline, wire it into the operating system (Windows first, cross-platform later), and leave me with clear documentation so the tool can be maintained and expanded. You’re free to pick the tech stack that suits you best; OpenCV, MediaPipe, TensorFlow, PyTorch, or a combination are all fine so long as the final executable runs locally without cloud dependence. Deliverables • A compiled desktop application (or installable Python bundle) that recognises at least five distinct hand gestures and maps them to system actions • A simple GUI for calibrating the camera, adjusting sensitivity and remapping gestures &bull...
...so the first step will be cleaning, transforming, and loading those structured datasets into an efficient pipeline. From there you’ll choose the most appropriate algorithms, iterate on feature engineering, and fine-tune the model until we reach an agreed-upon accuracy threshold. You should be fully comfortable with scikit-learn and pandas for data handling, and able to switch to TensorFlow or PyTorch if deep learning ends up adding value. Packaging the final model behind a lightweight API or another deployable artifact—so it can slot into our existing workflow—will be part of the scope. Deliverables: • Reproducible Python code (notebooks or modules) • Trained model file and versioned training script • Brief report explaining methodology, key...
...documentation and RTSP details will be shared after discussion. Demo Dashboard Requirements A simple web dashboard should include: Live Camera View AI Detection Overlay Event Alerts Event History Detection Screenshots Camera Management (Basic) Dashboard Statistics A polished UI is preferred but not mandatory for the demo. Preferred Technology Stack Python FastAPI YOLO OpenCV TensorFlow or PyTorch React.js PostgreSQL Docker Ubuntu Linux NVIDIA GPU Support (CUDA) Equivalent technologies are acceptable if performance and scalability are maintained. Future Scope If the demo is approved, the selected developer/team will continue with the complete platform development, including: Face Analytics Safety Analytics Vehicle Analytics Object Analytics Attendance Analytics Multi-loca...
...its predefined category with solid, measurable accuracy. Here is how I see the work unfolding: • Prepare and clean the image dataset, applying sensible augmentation so the network generalises well. • Select or design a CNN architecture (transfer-learning with ResNet, EfficientNet, or a custom model—whichever you believe will perform best) and implement it in Python using TensorFlow or PyTorch. • Train, tune hyper-parameters, and track performance with clear metrics; I want the training notebook or script fully reproducible on my side. • Deliver the trained model file, an inference script or API endpoint, and a brief report explaining your methodology, final accuracy, and any recommendations for future improvements. I will supply the images...
...with Docker and production deployment. * Experience integrating or managing GPU-powered backend services. * Strong debugging, architecture, and performance-optimization skills. * Excellent spoken and written English. * Ability to communicate technical decisions clearly. ## Preferred Experience Experience with any of the following will be highly valuable: * NVIDIA RTX GPUs. * CUDA, TensorRT, PyTorch, or GPU inference. * AI-powered text-to-3D or image-to-3D systems. * GPU rendering or mesh-processing services. * GLTF, GLB, OBJ, STL, or similar 3D formats. * Cloud GPU infrastructure. * Redis, BullMQ, RabbitMQ, or similar job-queue systems. * AWS, Google Cloud, Azure, RunPod, or another GPU platform. * SaaS authentication, subscriptions, cloud storage, and usage tracking. * Creati...
I’m building a computer-vision layer that can reliably recognise images and return structured results that the rest of my platform can consume. The core of the project is an AI-driven image-recognition model; I need someone who can take it from dataset preparation all the way to a callable endpoint. Here’s what success looks like for me: • A well-architected model (TensorFlow, PyTorch or comparable) trained on the dataset I’ll supply—or augmented with open-source sets you recommend—to reach an agreed accuracy threshold. • Clean, readable Python code with clear comments and a short README so future team members can retrain or fine-tune the model. • A lightweight API or Dockerised service that I can spin up on AWS/GCP and hit with a PO...
...long recordings distilled into fast-paced highlight reels. Each reel should automatically surface the biggest moments: clean knock-outs, jaw-dropping goals, buzzer-beater threes, and decisive submissions, stitched together with sleek transitions and scoreboard overlays. I’m looking for a solution that leans on AI rather than manual scrubbing. Whether you prefer computer-vision toolkits (OpenCV, PyTorch, TensorFlow) or proven AI editors like Wisecut, Descript, or run-time FFmpeg pipelines, I’m open, as long as the end product is: • A 60–120-second, 1080p (or better) reel for each event I supply • Accurate, exciting clip selection—no dull filler • Clean branding: my logo sting at open/close, lower-third for athlete names when visible &...
...concept to a defendable thesis. Your role will cover the critical stages: refining the research objectives, selecting or developing suitable AI models for power-system data, running the necessary simulations, and finally shaping the results into a clear, publication-ready document that meets university guidelines. Preferred tools are Matlab/Simulink or Python with libraries such as TensorFlow, PyTorch and Pandas, though I am open to alternatives if they better suit the chosen model or dataset. Deliverables • A detailed methodology outlining the AI techniques applied to the power-system problem • Simulation code and test data with clear commenting so I can reproduce the results on my end • Interim result summaries for each major milestone so feedback can be...
...quality—wrapped in a clean, browser-first interface that feels effortless for everyday users. Once that core experience is live and stable, I plan to extend the same platform to text-to-video, image-to-image, and image-to-video modules, so the architecture needs to be modular and GPU-scalable from day one. Here is the scope I need you to cover: • Develop the back-end inference service (Python, PyTorch or TensorFlow) and set up a queue system that can handle bursts of public traffic. • Create a lightweight React or similar front end with a single prompt box, style presets, and real-time generation status. No sign-in walls; usage should feel instant. • Integrate image safety filters and basic prompt moderation to keep the experience family-friendly. &bu...
...recommendation and ranking pipeline for several enterprise products and I need a senior-level engineer who already lives and breathes these systems. Your primary playground will be large-scale feature engineering and EDA, then building, tuning and shipping models that actually move CTR/CVR metrics in the wild. Here’s the environment you’ll be stepping into: Python and Spark for data prep, TensorFlow/PyTorch for model work, all running on AWS — S3, EMR and SageMaker. GenAI/LLM/RAG methods are in scope; I’m especially interested in seeing how you weave retrieval-augmented generation into ranking logic. You should be comfortable: • Designing and maintaining a feature store, then creating training/validation sets • Building custom neural networks a...
...polish of language plus a critical check of the scientific narrative, calculations, and reported metrics. No generative AI or automated writing tools may be used at any stage. Scope of work • Copy-edit grammar, spelling, and flow while retaining my academic voice. • Strengthen scientific explanations, notation, and figure/table captions where clarity is lacking. • Run the provided code (PyTorch + standard Python libraries), replicate the key experiments, and confirm that numbers in the draft match the reproduced outputs. • Flag any discrepancy in results or methodology and suggest precise fixes; inline comments in the PDF/LaTeX source are fine. • Deliver a clean submission-ready version plus one track-changes or commented file for my own learni...
I need an end-to-end AI automation solution built around unsupervised learning models. My main goal is to ingest raw data, discover hidden structures or groupings automatically, and trigger follow-up actions without manual oversight. Here’s the flow I’m envisioning: you help me prepare and normalise the...training, and inference • Documentation explaining the algorithm choice, hyper-parameters, and how to retrain with fresh data • A simple interface or notebook that visualises clusters, anomalies, or other discovered patterns • Deployment script or Dockerfile so I can spin everything up in one step If you have experience automating pipelines with tools such as scikit-learn, TensorFlow, PyTorch, or similar, and can balance model performance with...
...produce short, coherent passages consistent with the underlying signals in the data. You’ll take the raw vectors, handle any additional preprocessing, select suitable Natural Language Processing architectures—Transformers, LSTM hybrids, or anything you can justify—then train, validate, and fine-tune until performance is solid. Helpful toolkits are totally up to you; I’m comfortable with Python, PyTorch or TensorFlow, Hugging Face Transformers, and scikit-learn if they make the pipeline smoother. Deliverables I expect: • Jupyter notebooks or .py scripts with clear, reproducible code • Trained model weights and tokenizer files • A concise README explaining environment setup and how to run inference • Brief evaluation report showi...
...as complementary signals, learns their joint representation, and outputs a binary (disease / no-disease) or probabilistic risk score. Here is what I need from you. First, a clean data-pipeline that ingests my gene expression matrices alongside the associated clinical notes, handles any necessary tokenisation or normalisation, and keeps sample alignment intact. Second, a well-documented model—PyTorch or TensorFlow is fine—that includes separate encoders for each modality and a fusion layer able to capture cross-modal interactions before the final prediction head. Finally, solid training and evaluation scripts with clear metrics such as AUC, accuracy, precision-recall and, ideally, an ablation option so we can see the added value of each modality. Deliverables &bull...
...need a small, self-contained AI solution that can automatically create original, royalty-free, 15-second classical music reels ready for Instagram. The reels are meant purely for entertainment, so each piece should feel fresh, engaging, and unmistakably classical while still sounding modern enough to hold a scroller’s attention. Here’s what I’m after: • A model or pipeline (Python, TensorFlow/PyTorch, or a comparable framework) that composes and renders short classical-style tracks on demand. • Optional lightweight interface—command line or a simple web page—where I choose how many reels I want and instantly receive downloadable audio (or .mp4 with a static cover if that speeds posting). • Output that meets Instagram’s lengt...
I am looking exclusively for a local UAE-based developer...stakeholders can see live KPIs. • If source data is missing, the plug-in quietly launches a web-scraper (Python) to grab supplier specs and fills the gaps before updating Airtable. Acceptance criteria 1. Signed-installer (.msi or .exe) that adds a ribbon tab in Inventor 2024+ with the above workflow. 2. Python source, neatly documented, including the computer-vision model (TensorFlow or PyTorch—your choice) and the web-scraping routines (BeautifulSoup/Scrapy). 3. REST hooks or direct connectors for Airtable and Power BI proven to work with my API keys. 4. Short video demo plus a written deployment guide. If something in the toolchain needs tweaking, I’m open to suggestions—as long as Python...
...descriptions into full 2D animated videos—no manual key-framing required. The goal is an end-to-end pipeline that ingests text, generates the storyboard and assets, and stitches everything together into a smooth, short video (think explainer-style pieces up to a couple of minutes). Here is the scope I have in mind: • Core model: leverage or fine-tune existing diffusion/transformer frameworks (PyTorch or TensorFlow preferred) so the system can produce consistent characters and backgrounds frame-by-frame. • Animation logic: interpolate poses and lip-sync automatically; output 1080p, 24 fps MP4. • Interface: a lightweight web app where I can paste a script, set scene breaks, and press “Generate”. • Deliverables: clean, well-commented so...
...ur reply Do not waste your time. Be PRECISE OK? Scope • Cover the full spectrum of modern AI: classic machine-learning workflows, natural-language processing, computer vision, and emerging multimodal techniques. • Wherever a concrete example fits—image classification, text generation, or small end-to-end projects—walk through the code step by step. Feel free to showcase TensorFlow, PyTorch, Keras, or any other mainstream library if it helps keep explanations clear and practical. • Position every lesson for self-paced study: concise theory, a runnable notebook or script, annotated output, plus brief “Why this matters” commentary to keep motivation high. Deliverables 1. Structured tutorial content (written or slide deck) paire...
...strong experience in CUDA kernels, PyTorch, and performance optimization. Work setup: Remote work through AnyDesk or another remote desktop tool. You will work directly on my development environment. Tasks involve debugging, improving, and optimizing CUDA implementations. Each task typically requires around 3 hours of focused work. Payment: Rate: $8–$10 per hour depending on experience. Payment will be released after 2 weeks of completed work. This is intended to become a long-term working relationship with regular tasks. Required skills: Strong C++ programming skills. CUDA kernel development experience. Understanding of GPU memory optimization, CUDA streams, and kernel performance. Experience debugging CUDA correctness issues. Knowledge of PyTorch CUDA ...
...and optimising Python-based endpoints that expose LLM features to our web and mobile clients. You should feel at home writing clean, test-covered code in FastAPI (or a comparable framework) and managing everything that makes an API reliable—auth layers, rate-limiting, logging, CI/CD, and containerisation. Although the core models are LLMs, familiarity with common AI stacks such as TensorFlow, PyTorch or Scikit-Learn will help when we experiment with alternative architectures. Key deliverables • A version-controlled codebase in Python that wraps our existing LLM checkpoints behind REST (or gRPC) endpoints • Dockerfile and deployment scripts for staging and production • Unit and integration tests with ≥90 % coverage, plus concise Swagger/OpenAPI docs ...
I need a Senior Computer Vision and Deep Learning Engineer to build a complete production-r...edges. This is not a basic rembg installation and no hidden third-party paid API should be used. The developer will be responsible for the complete project, including AI model selection/fine-tuning, Python FastAPI backend, GPU optimization, frontend integration with my existing PHP/JavaScript website, testing, bug fixing, deployment and production launch. Required technologies include Python, PyTorch, OpenCV, image segmentation, alpha matting, ONNX/TensorRT, CUDA, Docker and GPU deployment. Complete source code, trained model weights, training scripts, deployment files and documentation must be handed over. Payment will be milestone-based after quality and speed testing against a privat...
...strong experience in CUDA kernels, PyTorch, and performance optimization. Work setup: Remote work through AnyDesk or another remote desktop tool. You will work directly on my development environment. Tasks involve debugging, improving, and optimizing CUDA implementations. Each task typically requires around 3 hours of focused work. Payment: Rate: $8–$10 per hour depending on experience. Payment will be released after 2 weeks of completed work. This is intended to become a long-term working relationship with regular tasks. Required skills: Strong C++ programming skills. CUDA kernel development experience. Understanding of GPU memory optimization, CUDA streams, and kernel performance. Experience debugging CUDA correctness issues. Knowledge of PyTorch CUDA ...
...行動電話, 險種名稱 and any other handwritten box on the page—that must be recognised automatically. Rather than hard-coding coordinates, I want to call the software with a list of field labels (for example “name” and “address”) and receive the recognised text for each label in a clean machine-readable format such as JSON or CSV. Internally you are free to use OpenCV, PaddleOCR, Tesseract, TensorFlow, PyTorch—or any combination—so long as the handwriting recognition works reliably on Chinese characters and can be extended if new labels are added later. Deliverables: • A runnable script or small service (Python preferred) that accepts a single image/PDF upload and returns the extracted values. • Clear instructions for installati...
...models, validate them against business KPIs, and work side-by-side with product and cloud engineers to deploy your work. Everything is green-field, so you’ll have freedom in approach provided you deliver clean, maintainable code. Must-haves • Deep command of Python for data processing, experimentation, and micro-service development (our core language). • Practical experience with TensorFlow, PyTorch, or a comparable framework, plus a solid grasp of feature engineering, experiment tracking, and MLOps principles. • Comfort with cloud services (AWS, GCP, or Azure) and containerisation so your models ship smoothly. Deliverables 1. End-to-end prototype of at least one core model, complete with well-documented Python code. 2. Concise technical write-up ...
...development—no data wrangling or system plumbing—so I expect clean, well-documented Python modules that import neatly and respect our current project structure. You’ll work against the latest commit in our private Git repo, build out the ML logic we’ve specced (classification and ranking models to start), and deliver passing unit tests so the CI pipeline stays green. Feel free to lean on scikit-learn, PyTorch, or TensorFlow if they keep things elegant and maintainable; the end goal is fast, accurate models that can be called from Dynamo’s core services with minimal overhead. Deliverables • Python source files implementing the requested algorithms • Corresponding unit/pytest coverage ≥ 90 % • Brief README explaining model assumptio...
...students who want to apply Python-driven machine learning to a genuine archaeological problem. The mission is clear: build an AI model capable of reading an undeciphered script, pushing the limits of CNNs and related deep-learning techniques. Your day-to-day focus will be on designing, training, and iterating the model—from data pipelines in Python through to hyper-parameter tuning in TensorFlow or PyTorch—until we reach publishable accuracy. For context, you’ll be working with: • Images of ancient scripts • Transcriptions of ancient texts • Annotated historical data Expect hands-on experimentation with OCR, image augmentation, and transfer learning, plus frequent reviews with archaeologists to validate outputs. I’ll provide mentori...
...(CSV/JSON) as the sole training source, complete with field explanations and sample records. I am open to whichever learning objective—classification, regression, or clustering—best suits the patterns you discover in the data, and I welcome your guidance on choosing the most effective approach. Deliverables I need from you: • A fully trained model in a portable format (TensorFlow SavedModel, PyTorch .pt, or ONNX). • A lightweight inference script or REST endpoint my app can call directly. • Clear documentation covering input schema, preprocessing steps, evaluation metrics, and future fine-tuning instructions. The surrounding pipeline and UI are already in place, so once the model and inference layer are ready, integration should be straightforw...
...reports, dashboards, or predictive models that guide business strategy. Here’s the current picture: • The priority is data analysis and insights, not simply model building for its own sake. • I have access to several structured and unstructured datasets; you can help determine which are most valuable and how best to preprocess them. • Flexibility around tools is important—Python, R, TensorFlow, PyTorch, scikit-learn, or cloud services such as AWS SageMaker and Google Cloud AI are all possibilities as long as the final solution is reproducible and well-documented. Key deliverables I’m looking for: 1. A clear, well-commented codebase or notebook that ingests the agreed data, cleans it, and applies the chosen algorithms. 2. An explanation (wr...
...detection • trend prediction • behavioral analysis • pattern recognition • forecasting ________________________________________ Dashboard Modern executive dashboard including: • KPIs • alerts • analytics • reporting • visualization • customizable widgets ________________________________________ Preferred Technology Backend • Python • FastAPI • PostgreSQL • Neo4j • Elasticsearch • Redis • Kafka AI • PyTorch • TensorFlow • LangChain • Llama • Vector Databases • RAG Frontend • React • TypeScript • Mapbox • D3.js or Infrastructure • Docker • Kubernetes • AWS or Azure • CI/CD ________________________________________ Deliverables &b...
...routine content updates or data-driven sections so the site stays fresh with minimal manual input. Platform choice is open. WordPress, Shopify, or a fully custom stack are all possible as long as the final build remains scalable, secure, and easy for a non-technical team to manage. Recommend whichever stack will integrate smoothly with services such as OpenAI, Dialogflow, or custom TensorFlow/PyTorch models. Deliverables 1. Wireframes and final responsive UI/UX. 2. Fully developed site with the three AI features above working end-to-end. 3. Admin dashboard and clear documentation for future maintenance. 4. Deployment to our hosting of choice plus a short hand-off session. Acceptance criteria • Pages load in under 2 s on mobile and desktop. • Chatbot repli...
... end-to-end computer-vision system that can spot diseases across three domains—human healthcare, crop agriculture, and veterinary practice—without ever calling external APIs. Scope • Training: You’ll design and train deep-learning models to recognise Cancer in medical scans, Plant Fungal Infections in leaf images, and Animal Parasitic Infections in animal skin images. Feel free to choose PyTorch, TensorFlow or another modern framework, but the workflow must run entirely on local hardware after setup. • Data: Combine public datasets and clear instructions for adding private data so future users can extend or fine-tune. • Inference: Produce a lightweight CLI (a small GUI is a bonus) that accepts single images or folders, returns class label...
...back-ends. Core responsibilities • Design and implement reproducible ML pipelines in Python. • Optimise training workflows for speed, cost, and accuracy. • Evaluate, version, and monitor models once they’re in production. • Collaborate with the cloud and full-stack teams to expose prediction services via robust APIs. Acceptance criteria • A functioning training pipeline (ideally using PyTorch or TensorFlow) that runs end-to-end on our AWS account. • Clear documentation and unit tests. • Demonstrable improvement over baseline benchmarks we will provide. If you also enjoy Rust, Rails, or architecting large-scale cloud infrastructures, there will be plenty of space to apply that expertise as the product grows. When you submit a ...
...through a lightweight FastAPI endpoint that I can drop into my existing stack. Here is what I need done: • Curate or expand an annotated dataset that covers cars, trucks, bikes, pedestrians, and common domestic or wild animals in varied lighting and weather conditions. • Train and fine-tune an object-detection model—YOLOv8, Faster-RCNN, or another state-of-the-art architecture in TensorFlow or PyTorch is fine as long as it balances accuracy and inference speed. • Validate the model with mAP, precision/recall, FPS, and confusion-matrix reports, then iterate until it meets the agreed thresholds. • Package the final weights and inference code behind a FastAPI REST endpoint that accepts an image file or base64 string and returns JSON bounding boxes,...
Job Description: I'm running a self-hosted, real-time conversational voice AI pipeline (STT → LLM → streaming TTS over WebSocket) on a RunPod GPU pod (CSM-1B TTS model, using the davidbrowne17/csm-streaming fork with Sesame's Mimi/moshi audio codec, PyTorch ). I need an experienced PyTorch/CUDA engineer to find and fix a reproducible latency bug that's blocking production readiness. The problem: At sentence/turn boundaries in a multi-turn conversation, generation stalls for ~7-8 seconds of dead air. I've already localized this precisely via wall-clock instrumentation: the stall occurs specifically inside Mimi's () call, when encoding a previous turn's generated audio into context — and specifically only w...
...Explore and implement Machine Learning techniques where applicable. Collaborate with domain experts to develop innovative software solutions. Participate in software testing, validation, documentation, and continuous improvement. Required Skills Strong proficiency in Python. Experience with NumPy, Pandas, and SciPy. Knowledge of Machine Learning using frameworks such as scikit-learn, TensorFlow, or PyTorch. Experience with SQL or other database technologies. Familiarity with Git and software development best practices. Strong analytical, mathematical, and problem-solving skills. Ability to understand technical concepts and translate them into software solutions. Preferred Skills Experience with optimization algorithms or scientific computing. Knowledge of data visualization tools...
...in mind 1. User snaps or imports a food image. 2. The app detects and segments each food item in the photo. 3. It matches each segment to a nutrition database and scales the values by estimated portion size. 4. Results are shown clearly on-screen and saved locally for later review. Technical direction • Target platform: Android only. • Feel free to leverage TensorFlow Lite, ML Kit, PyTorch Mobile or any other on-device vision stack you are comfortable with, as long as the APK remains lightweight and runs without cloud connectivity. • A small cloud fallback for edge cases is acceptable, but the core analysis should work offline. • Source code, model files, and a brief README explaining model training or fine-tuning steps form the final delivera...
...(LangGraph, CrewAI, AutoGen, OpenAI Agents SDK, or similar). >Experience building Generative AI and Retrieval-Augmented Generation (RAG) applications. >Fine-tuning and optimization of Large Language Models (LLMs). >Experience with vector databases such as Pinecone, Milvus, Weaviate, FAISS, or ChromaDB. >Knowledge of LangChain, LlamaIndex, and AI orchestration frameworks. >Strong understanding of PyTorch, Hugging Face Transformers, and model optimization techniques. >Experience deploying AI workloads on AWS (EKS, SageMaker, EC2, Bedrock, or equivalent cloud services). >Experience designing RESTful APIs, gRPC services, Docker, Kubernetes, and CI/CD pipelines. >Familiarity with cybersecurity concepts such as SIEM, SOC operations, threat intelligence, vulnerab...
...etc. 3. Output a GPS fix (lat/long) for every confirmed object. 4. Provide a lightweight runtime that can run either on the drone’s onboard computer or a nearby edge server with minimal latency. 5. Deliver documented test results on real or simulated runway scenes demonstrating the required accuracy in day and night scenarios. Deliverables • Source code and trained weights (TensorFlow, PyTorch or similar). • Deployment script or container. • Brief technical report summarising methodology, training data used, evaluation metrics and performance. Acceptance criteria When I feed the system a representative data set captured from the drone, it must detect ≥95 % of the target items with <5 % false alarms and return GPS coordinates within a m...
...Optimize Whisper or evaluate alternative multilingual ASR approaches * Build confidence scoring and error handling for low-confidence transcripts * Improve acoustic and linguistic feature extraction * Benchmark model performance using objective evaluation metrics * Document findings and recommendations * Work closely with the founding team on iterative experiments --- ## Required Skills * Python * PyTorch * Speech AI / ASR * Whisper or similar speech models * Hugging Face * Audio processing (Librosa, Torchaudio, etc.) * FastAPI * Git --- ## Preferred * Experience with multilingual ASR * Hindi/Hinglish speech processing * Code-switching research * Healthcare AI or cognitive assessment * Model evaluation and benchmarking --- ## Deliverables By the end of the engagement, th...
...development—no data wrangling or system plumbing—so I expect clean, well-documented Python modules that import neatly and respect our current project structure. You’ll work against the latest commit in our private Git repo, build out the ML logic we’ve specced (classification and ranking models to start), and deliver passing unit tests so the CI pipeline stays green. Feel free to lean on scikit-learn, PyTorch, or TensorFlow if they keep things elegant and maintainable; the end goal is fast, accurate models that can be called from Dynamo’s core services with minimal overhead. Deliverables • Python source files implementing the requested algorithms • Corresponding unit/pytest coverage ≥ 90 % • Brief README explaining model ass...
I’m building a responsive web app that runs smoothly on any iPhone browser and on Windows desktops. Its co...sub-5-second odds refresh and sub-60-second player/weather updates 3. Documented data sources, model methodology, and environment setup instructions 4. Admin panel for adding new bookmakers or tweaking model parameters 5. Automated tests covering data ingestion, model outputs, and UI critical paths I’m flexible on the final tech stack—React or Vue on the front end, Python (TensorFlow, PyTorch, or XGBoost) or a Node alternative for the AI layer are all fine as long as they deliver speed and transparency. Let me know how you’d structure the data feeds, what modelling approach you would use, and any previous work that proves you can bring real-ti...
...--- ## MLOps & Production Engineering Experience with: * Docker * Kubernetes * CI/CD pipelines * MLflow * Model versioning * Model deployment * Model monitoring * Experiment tracking * Feature stores * GPU inference optimization * REST APIs (FastAPI preferred) Cloud experience with at least one platform: * AWS * Google Cloud Platform * Microsoft Azure --- ## Nice to Have * Deep Learning (PyTorch or TensorFlow) * Fine-tuning LLMs (LoRA/QLoRA/PEFT) * Distributed training * Knowledge Graphs / GraphRAG * Computer Vision or NLP experience * Reinforcement Learning * Streaming data pipelines (Kafka, Pub/Sub) * Airflow or similar orchestration tools * Experience building AI agents for enterprise applications --- ## What We're Looking For The ideal candidate is: * Str...
...code needs to read the rule in , choose the right metric direction automatically and optimise accordingly. I picture the workflow like this: 1. The user drops the four task files into a fresh directory. 2. They run python solution.py. 3. The script detects the data type, the metric orientation and the task family, trains the best available model (scikit-learn, XGBoost, LightGBM, PyTorch, TensorFlow, Hugging Face transformers—use whatever fits), and writes ready for upload. Acceptance criteria • runs with `python3 ` on a clean machine (specify any pip installs). • The script finishes inside a sensible time window on a single GPU or CPU-only fallback. • It produces a numeric score close to the current public leaderboard baseline for
...Once the data layer is solid, the core requirement is a fault-detection model that can recognise patterns linked to hardware problems—temperature spikes, power irregularities, component degradation, and similar signatures. Accuracy, low latency, and clear interpretability matter more to me than cutting-edge algorithms for their own sake, so please choose whatever combination of Python, TensorFlow/PyTorch, or any proven libraries you prefer as long as it makes the results easy to explain and monitor. I also want the results surfaced through a lightweight, responsive website. Think dashboard rather than marketing page: live fault alerts, a historical view, and a simple way to download or stream the underlying data. Hosting can be on AWS or another mainstream cloud platform...
...and integrate AI capabilities into user-facing products - Improve system architecture, scalability, reliability, and performance - Collaborate on technical decisions and future AI product development Required Technical Skills - Strong experience with Python and AI/ML development - Experience with AI model training, fine-tuning, and evaluation - Knowledge of machine learning frameworks such as PyTorch, TensorFlow, or similar - Experience building AI pipelines, data processing workflows, and model integrations - Full-stack web development experience - Backend: Python frameworks (FastAPI, Django, Flask, etc.) - Frontend: React, , or modern JavaScript frameworks - Experience designing and integrating REST APIs - Understanding of databases, cloud infrastructure, and deployment workfl...
I’m refining an AI module that powers a heads-up, visually oriented interface. The core engine is Natural Language Processing, yet right now its intent detection and response generation drift off-target too often. I need your help to push raw and benchmarked accuracy noticeably higher without sacrificing latency. You’ll dive into the existing Python codebase (TensorFlow and a light PyTorch utility are already in play), audit the current model pipeline, then propose and implement improvements—be that better tokenisation, a more suitable transformer architecture, advanced data augmentation, or smarter post-processing. The interface overlays results on a visual HUD, so clean, deterministic outputs matter; hallucinations or fluffed confidence scores show up instantly ...
