// tharunkaarthik gopinath
CS Honours Co-op student with a Minor in Mathematics at University of Windsor, currently an IT Analyst Co-op at Entegrus. I build clean, performant software, from data pipelines and real-time AI dashboards to Chrome extensions with real users. Dean's Honor Roll, Presidential Scholarship, AWS certified.
Collaborative desktop word processor with live six-language translation, real-time multi-user editing over WebSockets, and Azure SQL persistence. Built in JavaFX with an embedded Quill.js rich text editor.
- Embedded Quill.js in a JavaFX WebView, bridging Java and JavaScript for rich-text editing, clipboard integration, and .docx export via Apache POI
- Engineered real-time multi-user editing over WebSockets with delta-based synchronization (~80ms latency), live cursors, user presence, and cross-network collaboration
- Integrated Azure Translator for translating six languages live, preserving formatting through delta-aware text batching and source restoration
- Secured document storage with Azure SQL-backed authentication, BCrypt password hashing, and JDBC prepared statements, so each user's documents persist across sessions, stay private to their account, and save automatically
Real-time LLM routing dashboard with SSE streaming. Visualizes prompt complexity, model selection, confidence, latency, and escalation decisions across a three-tier, heuristic-based routing system.
- Built an interactive LLM orchestration dashboard for a three-tier, heuristic-based routing system that scores prompts 0-100 and escalates to a stronger model when confidence falls below 0.75, surfacing each routing decision
- Implemented token-by-token response streaming via Server-Sent Events (SSE), rendering live routing metadata and latency metrics alongside model outputs
- Derived routing-efficiency metrics and model-tier distributions from structured FastAPI responses to quantify cost-quality tradeoffs in model selection
- Integrated three REST endpoints (standard, streaming, and analysis-only) with responsive loading, error, and retry states for asynchronous model execution
Chrome extension with 50+ installs and 5-star rating - injects reusable email templates into Gmail and Outlook using MutationObserver for SPA compatibility.
- MutationObserver for dynamic DOM updates in SPA environments
- Local persistence via Chrome Storage API with async CRUD operations
- Modular content script separating DOM injection from state management
Graphical Wordle recreation built in Python - six attempts to guess a five-letter word with color-coded tile and keyboard feedback.
- Full Tkinter GUI with centered window, letter input, and on-screen keyboard tracker
- Real-time tile updates with green/yellow/gray feedback validated against a 5-letter dictionary
- Win/loss screens with custom icon assets, Pillow image rendering, and Play Again flow
| proficient | PythonJavaJavaScriptCSQLHTML/CSSFastAPIReactJavaFXQuill.jsWebSocketsApache POIpandasNumPyREST APIsSSEGitLinuxPostgreSQLSQLiteSQL ServerAzureMicrosoft GraphOAuth 2.0MavenJiraAWSWindows ServerChrome APIOOPDSA |
| learning | Next.jsDocker |
- Engineered and deployed an unattended Python telecom expense pipeline using Microsoft Graph, Entra ID, Exchange RBAC, and SMTP; runs 3 times monthly, automating billing-cycle matching, reconciliation, and report delivery to save 2-3 hours of manual work per month
- Extended a Python/SQL Server ETL platform supporting ~70,000 utility meters and 4.5B+ historical readings by adding Honeywell meter-reading ingestion with source-data mapping, schema normalization, filtering, and validation
- Enhanced the customer-facing AI search by tuning crawler and content-ingestion configurations and evaluating retrieval relevance and generated answers
- Built a Genesys Cloud conversational AI outage-reporting workflow with NLU intents, AI-powered slots, speech-recognition tuning, OAuth 2.0, and Microsoft Graph email notifications
- Resolved 200+ enterprise IT tickets across system issues, network troubleshooting, and phishing investigations; supported organization-wide workshops on AI-assisted workflows
Dec 2026
- Led digital logic and computer architecture labs for 100+ students on Boolean algebra, sequential circuits, finite-state machines, and register-transfer logic
- Evaluated circuit designs and assignments, guiding students through logic optimization, state-transition analysis, and debugging
Apr 2026
- Represented 700+ students and gathered departmental feedback
- Co-organized WinHacks Hackathon and Git & LeetCode workshops
Oct 2025
- 89% Major Average · Dean's Honor Roll
- Presidential Level Entrance Scholarship
- AWS Certified Cloud Practitioner (Aug 2025)
Sep 2028
Led a 3-member team solving timed algorithmic challenges in Python. Placed 2nd among 27 UWindsor teams in the 2024 qualifiers and advanced to Regionals.
Competed in 4 national-level contests on number systems, logic circuits, and data structures. Ranked in the top 20% nationwide on the ACSL CS Programming contest, qualifying for the National All-Stars round as a finalist.
Open to co-op positions, interesting projects, and good conversations.
emailgopinatt@uwindsor.ca↗linkedinlinkedin.com/in/tharunkaarthik-gopinath↗githubgithub.com/tharungopinath↗