The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
A financial risk platform combining an LSTM neural network, SHAP explainability, and Llama 3 for automated compliance and risk scoring.
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Challenges
View all
Related posts
Well crafted, the details make this feel premium.
Admivo.in — Full-Stack Product
Built the complete Admivo.in ecosystem from scratch for a foreign education consultancy, covering the student-facing platform and internal operations. The system includes country/destination discovery, application journey, counsellor workflows, resource libraries, live admission updates, document management, sessions, and an admin dashboard for managing the entire ecosystem.
IMDB Movie Review Sentiment Analysis
Built an end-to-end Natural Language Processing system to classify 50,000 IMDB movie reviews as positive or negative.
The project includes a complete text preprocessing pipeline with tokenization, stop-word handling, POS-aware lemmatization, and negation preservation. I used TF-IDF vectorization with unigram and bigram features and trained and compared four machine learning models: Logistic Regression, SVM, Naive Bayes, and KNN.
The best-performing models achieved approximately 90% accuracy, with SVM and Logistic Regression producing the strongest results. I also built a Streamlit web application that allows users to enter reviews, select a trained model, and receive real-time sentiment predictions with confidence scores.
Key work:
Developed an end-to-end NLP preprocessing pipeline
Implemented TF-IDF feature extraction with n-grams
Trained and compared four classification algorithms
Evaluated models using accuracy, precision, recall, confusion matrices, and classification reports
Serialized trained models for deployment
Built a real-time Streamlit inference application
