Web Development Specialist
Developed responsive and secure client websites by integrating cutting-edge hosting solutions with custom front-end and back-end features. Ensured reliability and an enhanced user experience for a diverse, global clientele.
Web DevelopmentHostingUser ExperienceHTML CSS
Computer-Vision-Object-Identification
Designed and deployed a YOLOv8-based object detection system achieving 85% mAP@0.5 and 70% mAP@0.5–0.95. Exported the trained model to TorchScript for optimized CPU inference and deployed it with a user-focused Gradio interface on Hugging Face Spaces.
YOLOv8Object DetectionTorchScriptGradioHugging FaceComputer Vision
Hate Speech Prediction System
Designed a multimodal hate speech detection system combining Whisper for speech-text-to-text, Wav2Vec for audio embeddings, and BERT-based models for text analysis. Implemented early and late fusion strategies for robust classification of offensive content.
WhisperWav2VecBERTDockerHugging FaceMachine Learning
Shopify QA Chatbot
Built and deployed an AI-powered customer support chatbot using Groq's LLaMA-3 70B API and Gradio for an interactive web interface. Utilized Hugging Face’s datasets library to load Amazon QA data for contextual responses.
ChatbotLLMGroq APIGradioHugging FacePrompt EngineeringCustomer Support
Document Summarizer using RAG
Developed a Retrieval Augmented Generation based tool using BART-large-CNN, FAISS, and Sentence Transformers to process PDF/TXT/MD files into 150-token summaries. Built with Hugging Transformers, LangChain, Streamlit, and deployed via Docker on Hugging Face.
BART Large CNNFaissSentence TransformersHugging TransformersLang ChainStreamlitDocker
Recommender Systems
Implemented a movie recommender system from scratch using User, Item, and Content-Based approaches, achieving a precision score of 87%. Developed a movie recommender system leveraging User-Based Filtering and Item-based methods, integrating OMDB API to fetch movie posters in real time.
Recommendation SystemsUser Based FilteringItem Based FilteringContent Based FilteringOMDB APIPythonMachine Learning
Information Retrieval Model
Implemented a Boolean Retrieval Model using inverted and positional indexing, achieving approximately 98% accuracy on gold-standard query tests. Enhanced the system with a Vector Space Model using TF-IDF vectors and cosine similarity, achieving a precision of 84% and recall of around 96%.
Information RetrievalVector Space ModelTF IDFCosine SimilarityDockerHugging Face
Ontology Based HEC Requirement Assessment
Designed and implemented an ontology to model teacher qualifications, skills, and course requirements by manually extracting key features from transcripts. Utilized SPARQL queries to identify the most suitable teachers for courses such as NLP, Design Defects, and IS Audit & Control based on their academic backgrounds.
Ontology DesignSPARQLData ExtractionDatabase Management
Game Development Intern
Worked as a Game Development Intern for 3 months. Designed a game on Unity engine with C#.
Game DevelopmentUnityC#Game DesignSoftware Development