Published by
FICTA - 2023 [Springer LNCS]
Summary
Published research on a transformer-based approach for generating image descriptions, contributing to advancements in image understanding and natural language processing.
About
Highly accomplished Machine Learning Engineer with a Master's in Electrical Engineering and a proven track record in developing and deploying advanced AI/ML solutions. Expert in building scalable systems, optimizing models for performance, and leveraging deep learning frameworks to drive significant improvements in data processing, predictive accuracy, and strategic decision-making. Seeking to apply robust technical skills in Python, C++, Docker, and Kubernetes to innovate and deliver high-impact solutions in a dynamic technology environment.
Work
Los Angeles, CA, US
→
Summary
Currently developing and optimizing advanced AI/ML pipelines to enhance data processing, predictive analytics, and strategic decision-making for business intelligence applications.
Highlights
Built a retrieval-augmented semantic search pipeline using BERTopic and Latent Dirichlet Allocation (LDA) on 50K+ documents, achieving 85% coherence and boosting decision-making efficiency by 40%.
Engineered a transformer sequence model for automotive trend analysis, achieving 92% accuracy on time-series data and improving strategic decision-making by 30%.
Led development of a Keyword-Assisted Structured Topic Model in R, integrating 10K+ additional documents to boost topic extraction and contextual relevance by 15%.
Reduced query latency from 2.5s to 0.8s in Google BigQuery by optimizing partitioning and execution planning, enabling low-latency retrieval at scale.
Bhubaneshwar, Odisha, India
→
Summary
Developed and deployed robust ensemble forecasting models and optimized data pipelines to enhance financial predictions and streamline feature engineering for large-scale transaction records.
Highlights
Implemented an ensemble forecasting model (XGBoost and LSTM) for payment timelines, reaching 93% accuracy and outperforming baseline by 10% for reliable cash-flow predictions.
Engineered a feature pipeline with 32 novel predictors, boosting F1-score from 0.82 to 0.89 on financial time-series data.
Optimized SQL pipelines for 1.2M+ transaction records, cutting data preparation time by 35% and streamlining feature engineering.
Deployed ensemble forecasting models as RESTful microservices with Docker and Kubernetes, integrating real-time data aggregation to deliver scalable, low-latency inference in production.
Remote, US
→
Summary
Led a team in developing an e-commerce platform with integrated hybrid recommendation systems and architected low-latency ML microservices.
Highlights
Led a group of 3 to engineer an e-commerce platform with React, integrating a hybrid recommendation system (collaborative and content-based) to personalize product discovery.
Architected low-latency Flask microservices to deploy ML models, integrating C++ backends with Jenkins CI/CD, achieving sub 50ms response times at scale with real-time monitoring via Prometheus and Grafana.
Orchestrated large-scale A/B testing on personalized recommendation models, boosting click-through rate (CTR) by 12% and enhancing user engagement.
Education
Los Angeles, CA, United States of America
→
Master of Science
Electrical Engineering
Courses
Analysis of Algorithms
Computer Networking
Machine Learning
Multimedia Data Compression
Internet and Cloud Computing
Introduction to Digital Image Processing
Information Retrieval Systems and Web Search Engine
Skills
Programming Languages
Python, C, C++, SQL, Java.
ML/AI Frameworks
PyTorch, TensorFlow, Scikit-Learn, Hugging Face.
MLOps & Tools
Docker, Kubernetes, MLflow, Git, LangChain, LlamaIndex.
Cloud Services/Libraries
AWS (EC2, S3, SageMaker), Azure, Spark, PySpark.