Alina P.
AI and Data Science , Operations
(214) GO-RELAY / (214) 467-3529
Careers@RelayHumanCloud.com
Ahmedabad, India
SKILLS
- AI Proficient
- Agentic AI
- SQL
- Python
- PostgreSQL
- Github
- Java
- RAG
- FAISS
- Docker (Basic)
- HTML
- FastAPI
- Flask
- Streamlit
- Selenium
- Playwright
- Git
- Pandas
- NumPy
- OpenPyXL
- JSON
- CSV
- SQLite
- Scikit-learn
- TensorFlow
- API Development
- Web Automation
- AI Test Case Generation
- Data Analysis
- Logging
- Exception Handling
- Outlook API
- MSG Automation
EDUCATION
Master's in Computer Application
Vellore Institute of Technology, Chennai (2025)
Bachelor of Computer Application
Dr. Virendra Swaroop Institute of Computer Studies (2023)
ASSESSMENTS
PERSONALITY
ENFJ-A
People with ENFJ as a personality type are dedicated to positively influencing communities. They are thoughtful and idealistic, striving to contribute to others' well-being and uphold their values.
PROFESSIONAL SUMMARY
Detail-oriented Junior Software Developer with hands-on experience in Python automation, backend API development, and data-driven solutions. Strong expertise in Python, FastAPI, Pandas, and SQL, with exposure to machine learning workflows and process optimization. Proven ability to streamline operations through automation and build scalable backend applications.
PROFESSIONAL EXPERIENCE
Jr. Python Automation Engineer
SPR Consultech | India | May 2025 – Dec 2025
- Automated 8+ financial and operational reporting workflows using Python and Pandas, cutting report generation time from several hours to under 10 minutes per cycle.
- Built web and desktop automation pipelines with Playwright and PyAutoGUI, eliminating repetitive manual data-entry tasks for operations teams.
- Developed email and document extraction pipelines handling PDFs, Excel, and structured data, processing hundreds of documents per day with minimal manual intervention.
- Designed and deployed FastAPI REST APIs to integrate automation scripts with third-party business systems, enabling end-to-end process automation.
PROJECTS
Intelligent Document Processing Pipeline — Invoice & Receipt Automation
- Built an end-to-end serverless document processing pipeline using Azure Functions (blob trigger + HTTP triggers) to automatically classify, extract and analyze invoices and receipts uploaded by users.
- Implemented Azure Document Intelligence with three prebuilt models (prebuilt-layout, prebuilt-invoice, prebuilt-receipt) for intelligent document classification and structured data extraction with confidence scoring.
- Integrated GPT-4.1 Mini as both a classification fallback (when keyword scoring is inconclusive) and a receipt profiling agent that generates structured markdown reports with flags and warnings.
- Implemented P-Card compliance checks including high value detection, restricted merchant flagging, handwriting detection and payment method analysis for corporate finance teams.
- Built duplicate detection at two levels — filename check against processed container and invoice number check against CSV tracker — preventing double processing.
- Developed Azure AI Foundry Agent integration with two function tools (upload_document, get_results) allowing users to upload documents in chat and receive formatted results directly in the conversation.
- Applied security best practices including function key authentication, filename sanitization to prevent path traversal attacks, file size validation, and managed identity configuration for blob storage access.
Hybrid Live Data Chatbot (RAG + NLQ-to-SQL)
- Built an intelligent chatbot capable of answering both semantic (contextual) and quantitative (numerical) queries from the same interface.
- Implemented a RAG pipeline with LlamaIndex for document-grounded contextual responses, and a custom NLQ-to-SQL engine for precise aggregation queries (SUM, COUNT, filters).
- Designed a query classifier to dynamically route user questions to the appropriate engine, achieving accurate responses across both structured and unstructured data sources.
Intrusion Detection System for Internet of Vehicles
- Applied CNN architectures (VGG16, VGG19, YOLO) with TensorFlow to detect network anomalies in vehicular IoT traffic patterns.
- Performed systematic hyperparameter tuning to optimise model classification accuracy across multiple attack categories.
