Harsh Agrawal

Harsh Agrawal

Multimodal ML Researcher

👋🏻Hey there!

I’m Harsh Agrawal, a Machine Learning Engineer with a Master’s in Computer Science from Northeastern University. My expertise lies in Machine Learning and Natural Language Processing, and I thrive on building AI solutions that solve real-world challenges.

🧠 My Passion for AI and Machine Learning

My journey with computers began in 8th grade when I first started coding in Java. By the time I was in 10th grade, I had earned my certification and developed a deep fascination for the field of computer science. This passion led me to pursue a Bachelor’s degree in Computer Engineering, where I immersed myself in numerous courses, ultimately choosing to specialize in Machine Learning (ML).

📚 Educational Endeavors

My undergraduate years in Computer Engineering at Narsee Monjee Institute of Management Studies (NMIMS) in Mumbai were a transformative period during which I built a variety of ML projects and even published my research. This hard work culminated in multiple papers presented at prestigious IEEE conferences. Having recently completed my Master’s in Computer Science at Northeastern University, my focus remains on leveraging AI and ML to drive positive change by solving real-world problems. During graduate school, I honed my skills through advanced coursework in areas such as Programming Design Paradigms, Database Management Systems, Algorithms, Pattern Recognition & Computer Vision, and Machine Learning.

👨🏻‍💻 Professional Experience

Currently, I am a Machine Learning Engineer at BulkMagic, where I lead the development of AI-driven features for a group-buying platform. In this role, I built a real-time recommendation engine that boosted deal uptake by 25% during beta testing. I also optimized our data pipelines and dynamic pricing models, cutting processing latency by nearly 40% to enhance the platform’s scalability and user experience. Additionally, I’m exploring cutting-edge architectures (like transformer encoders and graph-based recommenders) to further personalize deals and drive user engagement.

In Fall 2023, I worked as a Data Science Co-op at Amazon Robotics, where I developed data-driven solutions to improve operational efficiency. I designed a downtime monitoring system for robotic arms using AWS Lambda and Athena, which identified the causes of 60% of downtime events and helped reduce downtime by 15%. I also built a model to automatically categorize support tickets by complexity using custom clustering algorithms, significantly streamlining the ticket triage process and reducing the backlog.

Prior to that, I served as an AI Developer at DosBro Infotech in Mumbai, where I spearheaded multiple AI projects. I engineered a content-summarization pipeline using transformer models like BERT and T5 for a major media platform, achieving a ROUGE-L score of 0.88 and accelerating news turnaround by 45%. Additionally, I developed a computer vision system with YOLOv4 to detect brand logos in social media images, attaining 89% accuracy and reducing manual tagging efforts by 40%

🌟 Project Highlights

  • Created an on-device disaster scene detection system, awarded 3rd prize at the 2nd Khoury College Project Pitch-a-thon.
  • Engineered and fine-tuned transformer-based language models to interpret clinical progress notes, achieving a Macro F1 score of 0.78 in a challenging healthcare NLP task.
  • Implemented a personalized GIF recommendation system using a multi-modal transformer model, achieving over 80% precision in suggesting contextually relevant replies.
  • Proposed a “One-vs-All” approach for classifying cassava leaf diseases, achieving 85.64% accuracy and publishing the findings in IEEE Xplore.
  • Designed a pneumonia detection system using image processing and deep learning, showcased at the 2021 IEEE International Conference on Artificial Intelligence and Smart Systems.

🚀 Future Aspirations

Looking ahead, I aim to stay at the forefront of Machine Learning and NLP, continually pushing the boundaries of these fields. I’m excited to take on new challenges and build innovative AI solutions that make a meaningful, positive impact on society.

📚 Sharing and Learning

I believe in the power of community and knowledge sharing. As a proud Kaggle Notebooks Expert, I enjoy mentoring others and contributing to the field of AI through collaborative learning and innovation.

🌌 Personal Interests

Beyond my professional endeavors, I am fascinated by the mysteries of space and love to unwind with music and gaming.

💌 Get in Touch

I am always excited to connect with like-minded professionals and discuss new opportunities. Feel free to contact me or connect with me on LinkedIn, GitHub, and Google Scholar. I look forward to hearing from you!