Machine Learning Data Science

DHRUV PRAJAPATI

SUMMARY Boston, MA Open to Relocating or Remote 210-***-**** *******@**.*** Github LinkedIn Impact-driven AI/ML Engineer specializing in transforming complex data into actionable strategies that drive business growth and efficiency. Proven expertise in developing scalable AI/ML solutions to optimize operations, uncover insights, and deliver measurable ROI. Skilled in aligning business needs with technical solutions, committed to continuous learning and delivering impactful results EDUCATION

MS in Computer Science, Boston University Expected January 2026 Coursework: Foundation of Machine Learning, Big Data Analytics, Advance Machine Learning and Neural Networks B.Tech in Computer Engineering, Mumbai University August 2019 - May 2023 Coursework: Data Structures and Algorithms, Data Warehousing and Mining, Applied Data Science, Natural Language Processing EXPERIENCE

Graduate Research Assistant August 2024 - December 2024 Boston University Boston, MA

• Designed and implemented the lightweight ggml library to enable efficient LLM inference on IoT devices, smartwatches, and smart- phones, achieving processing speeds under 10 ms per token

• Utilized model quantization techniques to optimize performance and reduce memory consumption, enabling seamless deployment of GPT-2 model on resource-constrained devices like smartwatch Machine Learning Engineer January 2023 - December 2023 Pask Info Tech Mumbai, India

• Architected an end-to-end transformer-based model for personalized ad recommendation system, optimizing the alignment between visual and textual features, which enhanced Visual Question Answering (VQA) task accuracy by 18%

• Orchestrated large-scale distributed LLM training, tuning, and tracking using Pytorch, HuggingFace, DeepSpeed, Ray and mlflow on AWS Sagemaker, achieving 8x faster training throughput and generating $50K monthly GPU savings

• Optimized hyperparameters of Vision Language Model (VLM) using a combination of manual and automated tuning techniques, which led to a 15% improvement in the recall rate of ad targeting

• Engineered RAG application with LLM using Langchain, Faiss and llama.cpp, boosting inference speed by 50% and reducing hallucina- tion rate by 10%

Data Scientist Intern May 2020 - August 2020

Prince Technologies Ltd Mumbai, India

• Engineered over 20+ user-centric, interactive dashboards using Google Analytics 4 (G4A) and Tableau to analyze and visualize website traffic, enhancing stakeholder understanding of key metrics increasing the revenue

• Deployed an AWS predictive analytics platform using Redshift and SageMaker for SVM and Random Forest, achieving 93% accuracy in estimating purchasing intension of a visitor on website

• Implemented an LSTM RNN model to predict shopping cart abandonment likelihood, resulting in a remarkable 37% increase in conver- sion rates and significantly boosting sales and user engagement

• Developed and automated an end-to-end ETL big data pipeline using AWS DMS, S3, Apache Spark, and Airflow to efficiently collect, clean, preprocess, and transform data from PostgreSQL into AWS Redshift, delivering 70% cost savings through an optimized ETL architecture

PROJECTS

StyleCLIP: Text-Driven Image Manipulation - (Python, NodeJS, Pytorch, GANs, Autoencoders, Prompt Engineering) — March 2024

• Architected an AI-driven web application that generates unique images based on user prompts, leveraging StyleGAN and the Contrastive Language-Image Pretraining (CLIP) model

• Refined the efficacy of image manipulation by 43% through optimized prompts, analyzing model biases & hyperparameter tuning Crime Alert Through Smart Surveillance - (OpenCV, Deep Learning, CNN) — March 2022

• Integrated time-series images into a ConvLSTM2D model that detected anomalies in streaming crime video; achieving over 75% ROC metric

SKILLS

Programming Languages C++, Java, Python, R, SQL, HiveQL, Libraries Numpy, Pandas, Tidyverse, ggplot2, OpenCV, Keras, TensorFlow, Pytorch, Scikit-learn Databases MySQL, PostgreSQL, MongoDB, Snowflake, AWS Redshift, Firebase, Cassandra Frameworks & Technologies TensorFlow, Keras, PyTorch, Django, Flask (RestAPI), Hadoop, Linux Tools Airflow, Pyspark, AWS(S3, Lambda, SageMaker, Kinesis), GCP Bucket, GCP BigQuery, Tableau, Power BI, JIRA, Git, Docker

Contact this candidate