Resume - Sahaj Singh

Education

University of Maryland College Park, Maryland

Concentration: Machine Learning
Related Coursework: Data Structures & Algorithms, Computer Organization & Programming, Machine Learning, Deep Learning, Artificial Intelligence, Statistics & Applications, Computer Networks.

Zasti Inc. Ashburn, Virginia

Built Python and FastAPI services for processing, indexing, and retrieving unstructured data through REST APIs using MongoDB Atlas, enabling fast, scalable contextual search across the application.
Developed a production conversational application integrating external APIs, contextual retrieval, structured data validation, and backend workflow orchestration, improving response relevance and reducing invalid or inconsistent outputs.
Containerized application services with Docker and created GitHub Actions CI/CD pipelines for automated builds, testing, and deployment to Google Cloud, reducing manual deployment effort and creating a faster, repeatable release process.

Infinite Computer Solutions Inc. Rockville, Maryland

Spearheaded automation of member rate code validation through Python scripting, eliminating manual errors and reclaiming 10 hours per week of team productivity.
Extracted, cleaned, and analyzed large CSV datasets using Python (pandas) and Excel, identifying and resolving member data discrepancies in collaboration with data analysts and operations teams.
Investigated root causes of data inconsistencies and implemented corrective workflows, reducing discrepancies by 30% and improving downstream machine learning model accuracy by 5%.

International Software Systems Inc. Greenbelt, Maryland

Implemented and maintained asset records in the IATS inventory system, improving device-tracking accuracy and reducing inventory discrepancies by 20%.
Executed comprehensive OS deployment across Windows, Linux-Ubuntu, and Kali platforms, resulting in 25% efficiency improvement through standardized infrastructure.
Optimized enterprise-wide hardware integration, delivering 98% network uptime and 15% performance improvement through strategic domain architecture.

Germantown, Maryland | Individual Project

Built a Python and FastAPI workflow orchestration service that coordinated specialized components for collecting preferences, allocating budgets, and generating travel itineraries.
Defined typed data contracts with Pydantic and implemented deterministic validation to prevent malformed data from passing between workflow stages. Integrated external services and structured error handling while separating model-generated output from deterministic business logic.

Ashburn, Virginia | Individual Project

Developed a Python image-classification pipeline for approximately 10,000 dermoscopic images, including preprocessing, normalization, augmentation, training, and evaluation.
Evaluated performance using accuracy, precision, recall, F1-score, confusion matrix, and ROC-AUC, achieving 90.9% test accuracy. Organized the training and evaluation workflow into reusable components for data loading, model execution, metrics generation, and experiment comparison.

Ashburn, Virginia | Individual Project

Designed an AI fitness chatbot leveraging the Meta Llama 2 model in Pytorch, Integrating RAG architecture with ChromaDB for knowledge retrieval.
Accelerated model performance through PEFT and LoRA techniques, achieving efficient deployment via 4-bit quantization.

Google Cloud Certified - Generative AI Leader Python Java C++ C

HTML/CSS FastAPI, REST APIs & Pydantic Google Cloud Docker, GitHub Actions & CI/CD MongoDB & vector search pandas & CSV data processing

PyTorch & TensorFlow scikit-learn LangChain & RAG Git & GitHub Visual Studio Code & Jupyter Notebook