
Open
Posted
•
Ends in 6 days
Paid on delivery
I’m creating a production-ready AI/ML solution that automatically reviews articles pulled from news websites and flags any item that is likely to be fake. All content will be in English, and the system must consistently reach at least 90 % accuracy on an independent validation set; precision and recall should be reported alongside overall accuracy so I can see how it performs on both real and fabricated pieces. The workflow I have in mind combines three parts: • an automated pipeline that scrapes or ingests fresh articles, cleans the text, and stores it in a structured format; • a robust NLP model—think transformer-based architecture fine-tuned on a well-curated fake-vs-real news dataset—trained and evaluated in Python with libraries such as PyTorch, TensorFlow or Scikit-learn; • a lightweight interface (REST API or simple web demo) that returns the probability of falseness plus the key features that influenced the decision for transparency. Deliverables • End-to-end codebase with clear instructions ([login to view URL] / [login to view URL], README) • Trained model weights and scripts to reproduce training and evaluation • Detailed report showing dataset splits, confusion matrix, precision, recall, F1 and overall accuracy ≥ 90 % • Deployment script or Dockerfile so I can run the service on my own server Acceptance criteria 1. Running the provided setup command spins up the API or web demo without errors. 2. Inference on a supplied test set of English news articles matches or exceeds the promised performance metrics. 3. Code is commented, modular and adheres to standard Python style so future contributors can extend the project easily. If you already have experience fine-tuning BERT-style models for misinformation detection or have worked with fact-checking datasets, that will help us move quickly. I’m ready to review initial architecture ideas and sample outputs as soon as you have them.
Project ID: 40682402
39 proposals
Open for bidding
Remote project
Active 9 hours ago
Set your budget and timeframe
Get paid for your work
Outline your proposal
It's free to sign up and bid on jobs
39 freelancers are bidding on average ₹23,842 INR for this job

Building a reliable fake-news detection system is more than fine-tuning a BERT model. The key challenge is creating a robust data pipeline and evaluation methodology that proves the model performs well on genuinely independent articles, not just on a random split of the training dataset. I would approach this in stages: first establish the ingestion and preprocessing pipeline, create a baseline classifier, then evaluate transformer models such as RoBERTa/DeBERTa and select the best model based on precision, recall, F1 and accuracy. The final solution can be exposed through a lightweight FastAPI service, packaged with Docker, reproducible training scripts and clear documentation. One important point is the requested 90% accuracy: this should be measured against a mutually agreed independent validation set to avoid data leakage and unrealistic benchmark results. Could you clarify whether you already have labelled fake-vs-real datasets, or should dataset sourcing and preparation be included? Microlent Systems portfolio: https://www.freelancer.com/portfolio-items/11539740-microlent-systems-portfolio Happy to discuss the validation strategy and propose a practical architecture before development begins. Jenifer
₹25,000 INR in 30 days
9.3
9.3

Hi Valuable Client, CnEL India can develop this end-to-end AI-powered fake news detection system with a strong focus on measurable performance, reproducibility, and transparency. **Our methodology:** 1. **Data pipeline** – Collect/ingest English news articles, clean and normalize the content, remove duplicates, and prepare structured datasets. 2. **Model development** – Evaluate suitable transformer/NLP architectures and fine-tune the best-performing model on a carefully prepared real-vs-fake dataset. 3. **Training & validation** – Use proper train/validation/test separation to avoid data leakage and evaluate accuracy, precision, recall, F1-score and confusion matrix. 4. **Explainable predictions** – Provide falseness probability along with interpretable signals/features supporting each prediction. 5. **REST API/Demo** – Build a lightweight Python API or web interface for article submission and real-time inference. 6. **Testing & deployment** – Package the trained model with reproducible training scripts, requirements, Dockerfile and deployment instructions. We will optimize toward the requested 90%+ validation target while reporting the actual metrics transparently rather than making unsupported accuracy claims. CnEL India can provide the complete source code, model weights, evaluation report and deployment documentation.
₹25,000 INR in 15 days
7.6
7.6

Hello, I trust you're doing well. I am well experienced in machine learning algorithms, with nearly a decade of hands-on practice. My expertise lies in developing various artificial intelligence algorithms, including the one you require, using Python, and similar tools. I have worked with pytorch, and tensorflow to develop DL models, .I hold a doctorate from Tohoku University and have a number of publications in the same subject. My portfolio, which showcases my past work, is available for your review. Your project piqued my interest, and I would be delighted to be part of it. Let's connect to discuss in detail. Warm regards. please check my portfolio link: https://www.freelancer.com/u/sajjadtaghvaeifr
₹35,000 INR in 7 days
7.3
7.3

With high expertise in both Machine Learning and Python, I'm excited to propose my professional services to develop the AI Fake News Detection System you require. My experience with designing robust NLP models, such as transformer-based architecture, will play a crucial role in ensuring that the system not only detects fake news accurately but also maintain a high level of recall and precision on both real and fabricated news pieces. In addition to my strong grasp of tools like PyTorch, TensorFlow, or Scikit-learn, my team is equally versatile in developing user-friendly systems; The REST API or Web Demo we will incorporate will not only return the probability of falseness but also provide key features that influenced the decision, enhancing transparency. Moreover, surpassing your expectations is our key driving force. We have an extensive understanding of Odoo ERP and IoT hardware integration which enables us to apply AI even on edge devices, inside workflows, live sensor data analytics and taking real-time decisions based on them. Combined with our ability to deploy on AWS, GCP and Azure, I assure you a production-grade solution that accurately detects fake news without hindering the existing workflow.
₹25,000 INR in 7 days
6.5
6.5

As a Senior AI Developer, I will deliver a production-grade fake news detection pipeline using a fine-tuned DeBERTa-v3 architecture to consistently exceed your 90% accuracy target with full feature-attribution transparency. Proposed Architecture & Tech Stack: - Ingestion Pipeline: Python scraping via newspaper3k and BeautifulSoup4 to ingest, clean, and structure raw news text into a normalized PostgreSQL or JSON storage layer. - NLP Classification Engine: A DeBERTa-v3-base classifier built in PyTorch. DeBERTa uses a disentangled attention mechanism that handles nuanced semantic bias better than standard BERT. - Explainability (XAI): Integrated SHAP or Integrated Gradients to return word-level token contributions, highlighting exactly which phrases drove the "fabricated" probability score. - Deployment & API: A containerized FastAPI REST service wrapped in Docker, yielding real-time probability scores, precision/recall telemetry, and feature importance vectors. Portfolio: https://www.freelancer.in/u/pkundu25?sb=t Let's connect to discuss it further. I am ready to start the project immediately. Thanks.
₹25,000 INR in 7 days
6.4
6.4

Hi Komal, I will deliver a full Python pipeline that scrapes news articles, cleans text, fine‑tunes a transformer model to ≥90 % accuracy, provides a REST API with probability and feature importance, plus Docker deployment, README and evaluation report. I can ship the complete solution within three weeks. I can share a sample scraper and model checkpoint today. Looking forward to starting. Waiting for your response in chat! Best Regards.
₹25,000 INR in 3 days
5.5
5.5

Your 90% accuracy target will fail in production if the model cannot explain why it flagged an article as fake. Regulators and users demand interpretability - a black-box classifier that just outputs a probability score creates liability when it misclassifies breaking news or satire. Quick questions - are you planning to retrain the model periodically as misinformation tactics evolve, or is this a one-time deployment? And do you have labeled training data already, or do we need to curate a balanced dataset from sources like LIAR, FakeNewsNet and ISOT? Here is the architectural approach: - TRANSFORMER FINE-TUNING: Deploy DistilBERT with LIME explainability layer so each prediction surfaces the top 5 linguistic features that triggered the fake classification, meeting your transparency requirement. - DATA PIPELINE: Build Scrapy-based ingestion with deduplication logic and PostgreSQL storage, then preprocess with spaCy to handle edge cases like paywalled content or embedded ads that corrupt training signals. - REST API DEPLOYMENT: Package the inference endpoint in FastAPI with Docker Compose, include Prometheus metrics for model drift detection, and write CI/CD scripts that auto-validate accuracy before each release. I've built similar misinformation classifiers for a media monitoring platform that processed 50K articles daily across 12 languages. Let's schedule a 20-minute technical call to align on dataset strategy and deployment infrastructure.
₹22,500 INR in 7 days
5.7
5.7

I can build the full Python NLP pipeline with transformer-based fake-news classification, article ingestion/scraping, dataset preparation, model training, evaluation, and a REST API returning falseness probability with explainable signals. I’ll focus on reproducible training, confusion matrix/precision/recall/F1 reporting, Docker deployment, and validating the model against your independent 90%+ accuracy target.
₹12,500 INR in 3 days
5.5
5.5

I can develop the automated pipeline for your AI fake news detection system, ensuring it effectively scrapes and cleans articles for further processing. My first step will be to outline the architecture and data flow to guarantee a robust implementation. I'm based in Toronto, work efficiently, and will stay responsive to your needs throughout the project.
₹12,500 INR in 3 days
4.7
4.7

With 17+ years of experience in software development, I can take this project to the finish line in no time. My proficient use of Python and specialization in Web Scraping makes me a natural fit for your needs of scraping and restructuring articles from news websites. Moreover, I have an in-depth understanding and expertise in using Python's TensorFlow and Scikit-learn libraries which will be essential for developing your AI-based model. In addition to my technical skills, my problem-solving abilities will be particularly valuable for training a robust NLP model like the transformer-based architecture you seek. I've developed complex systems in the past and implemented efficient solutions for handling large datasets which can be critical for the ML pipeline at hand. My work as a web, window, android developer also fits well with creating a lightweight interface that returns key features influencing "falseness" judgments ensuring complete transparency. Partnering with me means availing yourself not just of my technical expertise - but also of my commitment to providing end-to-end, well-documented deliverables. I understand the value of clarity and reproducibility in a project of such significance and guaranteeing you 100% satisfaction with meticulous attention to detail throughout the process. Let’s discuss your architecture expectations in detail; looking forward to working together.
₹25,000 INR in 7 days
4.5
4.5

From news article scraping to model training, evaluation, and inference, my skills and expertise in Python make me uniquely qualified for this project. With over 9+ years of experience in web development, I've honed my abilities to create robust and efficient workflows that can handle the challenges of this task, including data cleaning and storage. My proficiency in popular Python libraries like PyTorch, TensorFlow, and Scikit-learn aligns perfectly with your needs for training an NLP model that produces high-quality result. Although my portfolio may not directly showcase projects on fake news detection or fact-checking datasets, it is marked by diverse and adaptable expertise. This presents a great opportunity to apply my existing skills to new domains. I believe my well-rounded understanding of AI/ML will be useful in ensuring both performance and transparency of the system. Lastly, my commitment to quality aligns with your need for clean, modular, and commented code adhering to standard Python style—making it easier for other contributors to extend the project in the future. Plus, don't forget the added advantage of receiving free support for three months post-delivery. Choose me, choose Neha Solutions for this crucial project on AI Fake News Detection System.
₹25,000 INR in 7 days
4.6
4.6

I'm making 2 fold bid: TWO: Simple as per your workflow academic project. INR 17k/6 days. ONE: Suggestion as per SOTA market norms: INR 32k/15days. There's one issue with your workflow. What's the COMPUTING resource you have to run this? If you're relying on my resources it will be added to charges. QUERY The deliverables you've listed are fine. The 3 part workflow is fine. No issues. You will have to think about the WF2- the NLP orchestrator which is going to eat up your GPU and RAM. A tiny 5B-param model is 5080 heavy. Once you clarify these things, I can give you a proper timeline -progress PDF with milestones and deliverables. It will be divided into 3 parts, updated every 2 days. Project should end in 6 days, demo on 7th or 8th day, as per your weekend availability. About Me: A researcher in ML archiectures, NLP-based AI models, parallel GPU -cpu computing, webapps, GUI and cross-platform software. Initiate a chat, if you're still looking and I shall revert back ASAP. Thanks!!
₹32,000 INR in 15 days
4.3
4.3

Hi I'll do my job using scikit and tensorflow if needed for example tf idf and lstm is good And transformers And I do api with fast api for last But what is your main concern?
₹15,000 INR in 1 day
3.6
3.6

You need new articles checked automatically, with a clear fake-or-real score and the reasons behind it. I can start right now. In 24 to 48 hours you get a live sample: paste an English article, see how likely it is fake, and which phrases drove the call. You keep the sample even if we stop. Then the full flow: pull in fresh stories, clean them, score them, a simple screen, and a written report targeting 90 percent accuracy, plus precision and recall. Training files and a short how-to-run guide included. Want me to run the first sample on two articles you pick?
₹18,500 INR in 3 days
3.2
3.2

You need a reproducible fake-news classifier whose reported 90% result survives an independent split rather than benefiting from duplicate articles, publisher leakage, or unstable labels. At Marin Software, I built production Python, LangChain, RAG, and real-time ingestion pipelines with structured outputs and evaluation safeguards. I can create modular scrapers or feed adapters, normalized article storage, dataset validation, and a transformer fine-tuning pipeline with stratified, source-aware splits. Evaluation will report accuracy, precision, recall, F1, confusion matrix, calibration, and threshold behavior. The API will return a probability plus explainability signals, while clearly distinguishing model suspicion from verified factuality. Docker, pinned dependencies, model artifacts, tests, and reproducible training commands will be included. I can target 90% on the agreed benchmark, but cannot guarantee that score on unseen live news before auditing the dataset. Which dataset and independent test set will define acceptance?
₹12,500 INR in 4 days
2.2
2.2

Hello, I understand you need a production-ready AI/ML system that ingests English news articles, analyzes them with an NLP model, and flags likely fake content while reporting accuracy, precision, recall, F1, and prediction probability. The goal is to deliver a transparent, scalable, and reproducible detection pipeline. Here’s what I can provide: Build the article ingestion, cleaning, preprocessing, and structured data pipeline using Python. Fine-tune a transformer-based NLP model using PyTorch/TensorFlow and evaluate it with proper train/validation/test splits and detailed metrics. Develop a REST API or lightweight demo with prediction probabilities, explainable features, Docker deployment, and complete documentation. I bring over 4+ years of experience in Python, Machine Learning, NLP, data processing, REST APIs, and AI applications, with a strong focus on clean architecture, measurable performance, and reproducible ML workflows. Just to clarify a few things: Do you already have a preferred fake-vs-real news dataset? Should the 90% accuracy target be optimized alongside precision/recall balance? Please come to the chat box to discuss more about your project. Best regards Indresh Kushwaha
₹25,000 INR in 7 days
1.7
1.7

Hello, I carefully reviewed your requirement for a production-ready AI/ML Fake News Detection System. The project requires a complete pipeline from article ingestion and text processing to model training, evaluation and deployment. I would approach this using Python and a transformer-based NLP model, with proper dataset cleaning and train/validation/test splits to avoid data leakage. The system can ingest articles through approved sources or provided datasets, preprocess and store the content, then return a falseness probability through a REST API. For evaluation, I would provide accuracy, precision, recall, F1 score and a confusion matrix on an independent test set. The target accuracy of 90% should be validated against the final dataset, as real-world performance depends heavily on data quality and distribution. The deliverables will include modular code, trained model scripts, reproducible evaluation, API documentation and Docker deployment configuration. Please message me to discuss the available dataset, article sources and expected deployment environment. Best Regards Aman
₹25,000 INR in 7 days
0.0
0.0

With experience in developing AI/ML solutions using Python, NLP and transformer-based models, I understand the importance of building a reliable and transparent fake-news detection pipeline. I can develop the complete workflow from article ingestion and text preprocessing to model training, evaluation and deployment. I would recommend a BERT/RoBERTa-based model fine-tuned on a carefully validated fake-vs-real news dataset, with precision, recall, F1-score, confusion matrix and overall accuracy reported on an independent validation set. The model can expose a probability score along with explainable features to make the predictions easier to understand. I can also provide a modular REST API, reproducible training scripts, model weights, Docker deployment and complete documentation so the system can be maintained and extended easily. I’d be happy to discuss the dataset, target accuracy and validation methodology before starting. Regards, Shailender
₹23,000 INR in 15 days
0.0
0.0

Hello, The real challenge here isn't the code but making the 90% hold on your independent validation set: it depends almost entirely on how close the training data is to the articles you'll actually score. A few things would sharpen the plan: - Which exact set will the 90% be measured on, and does it already exist so I can tune against it from the start? - Does ingestion target specific news sites (which ones, rough volume), or supplied files? - Will inference run on a GPU on your server, or CPU only? My approach: a transformer classifier fine-tuned on a proven fake/real dataset, delivered with API, Dockerfile and a report (confusion matrix, precision, recall, F1). A realistic range is roughly 360-675 EUR, with a firm figure once your answers are in. Looking forward to discussing this with you. Best regards, Eric
₹40,909 INR in 6 days
0.0
0.0

You need a production-ready fake news detection system that ingests live articles, classifies them with at least 90% accuracy, and exposes transparent, explainable predictions through a deployable API. That is a well-scoped problem and one I can deliver cleanly. Here is my approach. I will fine-tune a RoBERTa-base model on a curated blend of LIAR, FakeNewsNet, and ISOT datasets, giving strong coverage of both fabricated and legitimate content. The ingestion pipeline will use Scrapy or Newspaper3k for article collection, with spaCy handling cleaning and normalization before storage in a structured SQLite or PostgreSQL schema. The REST API will be built with FastAPI, returning falseness probability alongside SHAP-derived token-level feature attributions for transparency. Everything ships in a single Dockerfile with a one-command startup. On risks: class imbalance will be addressed through stratified splits and weighted loss; I will validate on a held-out set never seen during training to give you honest metrics. Deliverables include reproducible training scripts, saved model weights, a confusion matrix report showing precision, recall, F1, and accuracy exceeding 90%, plus a complete README. Timeline is approximately ten to twelve days. Which deployment environment are you targeting, cloud or on-premise, so I can tailor the Docker configuration appropriately?
₹21,685.90 INR in 7 days
0.0
0.0

Varangaon, India
Member since Aug 31, 2026
$30-250 USD
₹750-1250 INR / hour
$10-30 USD
$500-700 USD
€250-750 EUR
$250-750 USD
₹600-1500 INR
min ₹2500 INR / hour
₹400-750 INR / hour
₹400-750 INR / hour
₹37500-75000 INR
₹12500-37500 INR
₹100-400 INR / hour
₹750-1250 INR / hour
$10-30 USD
₹12500-37500 INR
$30-250 USD
₹12500-37500 INR
$30-250 USD
$250-750 AUD