Syed Amaan's Projects
Speech recognition: Sequential MFCC distinguishes accents with 95% accuracy using various classifiers.
Automatic Speech Recognition system using transformers for real-time transcription.
PyTorch bigram character model for text generation, using gradient descent on Shakespeare corpus
Track and manage budgets efficiently with the Budget Manager Python module.
Scalable LLM deployment with retrieval-based systems, microservices, LangChain dialog management, and model productionalization.
Code files and my notes from deeplearning.ai's "Building AI Applications Using ChatGPT API" by Andrew Ng
the AI-native open-source embedding database
Vigenère cipher for text encryption, decryption, custom key usage, classical cryptography, alphabetical offset, secure communication.
Solidity smart contract 'CrowdFunding' for Ethereum, allows contributions, manager-controlled fund access.
This project scrapes, cleans, analyzes, and predicts cryptocurrency trends using historical data from CoinMarketCap
Image classification with CNNs on CIFAR-10 using TensorFlow, Keras, and data augmentation.
My notes and notebooks from DeepLearning.AI's short courses, updated as I finish each new course.
LSTM neural nets on agriculture data, a user-friendly web interface for CSV file uploads, results visualization, and performance evaluation
OrderBot: Customizable Python chatbot using GPT for restaurant order handling and delivery.
web app using JavaScript, Fetch API for latest posts with asynchronous data fetching, dynamic HTML generation, and date formatting.
GPT-style model using transformers, multi-head self-attention, dropout, and layer normalization.
Gradio-powered AI tools for interactive web applications
Implementation and visualization of diffusion models for image generation using PyTorch and U-Net architecture.
Collection of LeetCode questions to ace the coding interview! - Created using [LeetHub](https://github.com/QasimWani/LeetHub)
LLM training in simple, raw C/CUDA
Facial recognition: SRGANs enhance low-res images, featuring MTCNN, Siamese Model.
Python credit card validator using checksum, digit manipulation, even-odd separation, and string translation for secure number verification.
Manually backpropagating and optimizing a multilayer perceptron with batch normalization in plain Python (without PyTorch)
A tiny scalar-valued autograd engine and a neural net library on top of it with PyTorch-like API
NLP with RNNs: Sentiment analysis, character generation using LSTM, Python, TensorFlow.
Neural Networks: Zero to Hero
Secure password generator in Python with `secrets`, customizable length, numerics, specials, uppercase, and lowercase enforcement.
Reinforcement learning: Q-Learning agent optimized for navigating OpenAI Gym's Frozen Lake.
RAG app with Llama3, OpenAI embeddings, and ChromaDB for querying PDFs (adaptable for CSVs, JSONs and more)