Hi, I’m Eesha Khan, a PEC Level 2 Software Engineer specializing in Machine Learning, AI, and Data Science. I build intelligent systems and data-driven solutions that address complex, real-world challenges, including predictive models, AI-powered applications, and scalable data pipelines.I actively engage in hands-on projects and competitive platforms, exploring new technologies to enhance the effectiveness and reliability of AI systems. My goal is to contribute to innovative solutions that combine technical rigor with practical impact, while continuously growing in the evolving AI landscape.
Deep Learning & NLP: BERT, DeBERTa, CNNs, RNNs
LLMs & GenAI: LangChain, Ollama, Groq API
Problem: Users cannot reliably query and extract information from large PDF documents using standard chatbots.
Approach: Applied Retrieval-Augmented Generation to retrieve and generate relevant document context.
Outcome: Enabled accurate, context-aware question answering over uploaded documents.
Link: https://github.com/EngrEeshaKhan/rag-chatbot
Problem: Writing professional outreach emails is time-consuming for job seekers and freelancers.
Approach: Used a large language model to generate structured cold emails from user input.
Outcome: Reduced email drafting time while improving message clarity and professionalism.
Link: https://github.com/EngrEeshaKhan/AI-Powered-Cold-Email-Generator-Job-Client-Outreach
Problem: Early indicators of problematic internet use in children are difficult to detect manually.
Approach: Built a predictive model using behavioral and physical activity data.
Outcome: Supported early identification and intervention for healthier digital habits.
Link: https://github.com/EngrEeshaKhan/Child-Mind-Institute-Problematic-Internet-Use
Problem: Manual identification of protein complexes in cryo-electron tomography data is slow and inefficient.
Approach: Applied deep learning models to classify protein structures from 3D tomographic data.
Outcome: Enabled scalable and automated biological structure analysis.
Link: https://github.com/EngrEeshaKhan/CZII-CryoET-Object-Identification
Problem: Early melanoma detection from skin images is challenging due to subtle visual differences.
Approach: Applied DIP and handcrafted feature extraction followed by ML classification.
Outcome: Improved accuracy and reliability of melanoma detection.
Link: https://github.com/EngrEeshaKhan/Automatic-Melanoma-Detection-using-Hybrid-Features-and-Machine-Learning-Models
Problem: Manual essay grading is time-consuming and inconsistent.
Approach: Used NLP-based models to evaluate essays based on structure, coherence, and semantic quality.
Outcome: Produced automated scores closely aligned with human evaluation.
Link: https://github.com/EngrEeshaKhan/Learning-Agency-Lab---Automated-Essay-Scoring-2.0