Sruthisri Venkateswaran

- Built a production-grade RAG system powering an academic advisor chatbot, answering questions on courses, faculty, and research areas with 90%+ accuracy. - Architected dual RAG pipelines: persistent Qdrant-backed solution for structured corpora and lightweight in-memory pipeline for rapid experimentation - Engineered scalable web crawling and ingestion workflows using Playwright and BeautifulSoup, processing 100+ web pages with domain-restricted BFS routing - Automated transformation of unstructured content (PDFs, HTML, policies) into clean Markdown, removing boilerplate and navigation artifacts - Designed modular indexing pipeline with LlamaIndex, implementing hybrid dense + BM25 retrieval for improved answer completeness on aggregate queries - Integrated Google Gemini (gemini-2.5-flash) with conversation memory and query condensation, enabling robust multi-turn interactions - Established reproducible end-to-end workflows for data acquisition, indexing, and runtime, enabling seamless extensibility for ML and backend teams

- Developed computer vision pipelines for parking availability detection, achieving 92%+ accuracy across varied lighting and occlusion scenarios - Led training of YOLO-based models on 10,000+ annotated images, reducing detection latency by 20% - Applied SRGAN and InvSR super-resolution techniques, improving low-resolution input quality by 40% in PSNR scores - Built modular real-time analytics pipeline supporting 60 FPS video processing with live parking heatmaps and occupancy visualizations - Accelerated model training and inference using CUDA-enabled GPUs, cutting training time by 35% - Contributed to deployment planning for city-scale smart parking network monitoring 1,000+ spaces

- Spearheaded refactoring of Python-based geospatial simulation tool, reducing code complexity by 40% and improving runtime efficiency by 30% - Re-architected I/O subsystem, replacing legacy .mat dependencies with CSV and NetCDF support, increasing cross-platform data compatibility by 100% - Designed CLI and GUI interfaces, cutting simulation setup time for non-technical users by 50% - Translated 5,000+ lines of MATLAB algorithms into scalable Python using NumPy, Pandas, and GDAL, boosting processing speed by 35% - Authored 40+ pages of technical documentation and automated packaging workflows for open-source readiness and reusability - Reduced manual configuration errors by 80%, saving ~5 hours/week in team time

- Built and maintained 15+ React Native UI components, reducing mobile load times by 25% - Integrated Google and Apple OAuth authentication, decreasing user onboarding drop-off by 40% - Connected frontend with AWS-hosted REST APIs and ConnectyCube, enabling real-time messaging for 500+ concurrent users - Led cross-platform integration (React Native, MySQL, AWS), ensuring 100% data sync accuracy - Maintained CI/CD pipelines, reducing app deployment time from 1 day to under 2 hours

- Collaborated with 5+ cross-functional teams to deliver scalable ML and analytics solutions impacting millions of retail customers - Processed petabyte-scale datasets using SQL, Python, and PySpark across Databricks, Snowflake, and SAP HANA, reducing processing time by 50% - Designed advanced recommendation systems (collaborative filtering, hybrid models) with real-time Kafka integration, increasing user engagement by 25% - Led NLP-driven sentiment analysis on 500K+ product reviews, improving campaign targeting accuracy by 30% - Built 15+ Tableau dashboards and 5+ data marts, reducing stakeholder decision-making time by 30%

- Developed RESTful APIs for role-based ticket booking system, handling 10,000+ daily transactions with 99.9% uptime - Created optimized MySQL stored procedures, reducing query execution time by 40% - Integrated backend APIs with frontend modules, enabling real-time status updates with sub-second latency - Conducted 100+ Postman test cases, eliminating 95% of critical bugs pre-deployment - Documented 20+ API endpoints, reducing developer onboarding time by 60%

- Coursework: CSE 574: Machine Learning CSE 587: Data Intensive Computing CSE 565: Computer Security CSE 531: Algorithms Analysis and Design CSE 676: Deep Learning CSE 521: Operating Systems CSE 560: Data Models and Query Language CSE 611 Project Development - Won 2nd place in CSE Demo Day Spring 2025 for Full Stack Development Project on Qu Anytime.