Karthikeyan Balachandran

RTA-GPT: Enterprise AI & Call Center Intelligence Platform (Flagship Project)
Tech Stack: Python, FastAPI, LangChain, ChromaDB, Supabase, Qwen2.5-7B, GPT-4 Turbo, Whisper, YOLOv8, Streamlit, Ollama, Docker, Vercel

• Architected and delivered a comprehensive AI-powered Call Center Intelligence Platform that transforms RTA’s citizen engagement operations through Generative AI, multilingual understanding, and real-time analytics.
• Built an enterprise-grade RTA-GPT platform—a ChatGPT-like assistant trained on RTA data—to answer public queries across Nol, Salik, Licensing, Parking, Fines, and Metro with domain-specific precision.
• Designed a Retrieval-Augmented Generation (RAG) pipeline enabling English and Arabic Q&A via vector databases (ChromaDB, Supabase) for instant document retrieval and contextual LLM reasoning.
• Engineered real-time sentiment analysis dashboards that analyze thousands of daily citizen calls and chats, identifying tone, intent, and satisfaction levels with 92% accuracy.
• Developed an AI complaint triage engine that automatically classifies call topics and routes cases to specialized departmental agents (“DepartmentGPT”), reducing manual workload by 80%.
• Integrated speech-to-text (Whisper) and emotion detection to evaluate live audio interactions, providing supervisors with actionable insights to improve training and performance.
• Implemented an AI governance framework ensuring ethical, transparent, and compliant LLM deployment in alignment with UAE Smart Government data standards.
• Led the end-to-end AI lifecycle—data ingestion → embeddings → retrieval → model orchestration → visualization—delivering scalable multimodal (text, audio, video) pipelines.

Complementary AI Systems Integrated into the Platform:

• AI-Generated BRD Assistant: Automated project documentation, cutting preparation time by 65%.
• Goal-to-Task Automation Engine: Converted user goals into executable workflows via LLM reasoning.
• Traffic AI System: Deployed YOLOv8-based vehicle detection and signal optimization, improving traffic flow by 25%.
• User Flow Generator: Translated textual requirements into visual process diagrams using spaCy + Graphviz.

Impact:
Delivered Dubai’s first end-to-end Call Center AI ecosystem, combining Generative AI, NLP, RAG, and analytics to revolutionize citizen service delivery.
Resulted in 2.5× faster query resolution, 80% reduction in manual categorization, and 40% improvement in customer satisfaction—setting the standard for public-sector AI adoption in the UAE.

Roles and Responsibilities:

• Provided complete end-to-end solutions including requirement gathering, architectural design, system implementation, and deployment strategies.
• Solely managed code reviews, release processes, and deployments using GIT and automation tools, ensuring smooth delivery cycles.
• Managed and mentored both offshore teams from India and internal team members, fostering positive collaboration that improved productivity and on-time delivery.
• Guided development teams throughout the SDLC, taking full ownership from design to production deployment.
• Delivered technical guidance and support to external department teams, particularly for the Happiness Meter project involving feature enhancements and code fixes.

Major Accomplishments:

• Migrated legacy RTA systems to a modern Spring/Hibernate architecture, improving performance and scalability.
• Led the complete revamp of the RTA Portal , integrating it with Identity Access Management (IAM) for centralized, secure access.
• Developed and deployed integrations with external government systems like Salik , Licensing , Nol Plus , GIS , and more.
• Delivered multiple critical modules for Nol , Parking , and Public Transport systems (Marine, Abra, Naqel).
• Built custom, role-based dashboards and reports for several RTA departments to improve visibility and decision-making.
• Played a key role in the RTA 360 Vision project, successfully delivering all modules within tight deadlines and zero major defects.

Major Reusable Component Development:

• Developed a Call Center dashboard for the Customer Happiness Center to search, update, and generate user reports.
• Built a Finance Team dashboard for managing and downloading payment transaction records.
• Designed and implemented a reusable module framework for Dubai Smart Government (DSG) payment systems , enabling new module delivery within a day.
• Created a shared utility module to manage common functions, significantly reducing code duplication and improving maintainability.
• Built a dynamic form system with configurable fields and database design to support award applications like DAST , Driver Delivery Award , and RTA Scholarship .

Major Accomplishments:

• Designed and implemented build automation using Maven and Hudson , streamlining the CI/CD pipeline from development to production and significantly reducing deployment time.
• Independently developed and deployed critical modules such as Ask the Expert , Car-Free Day , and Survey , handling everything from backend logic and frontend UI to database schema modeling.
• Architected and implemented Microservices architecture to expose secure and scalable REST APIs for stakeholders including internal departments and third-party vendors.
• Applied microservice principles such as loose coupling, independent deployment, containerization, and domain-driven design (DDD) to ensure high availability and easy maintenance.
• Defined inter-service communication protocols using REST/JSON and built stateless services with proper logging, error handling, and monitoring metrics for improved observability.

Major Accomplishments:

• Developed full-featured eCommerce websites for top Al-Futtaim retail brands including ACE , Plug Ins , and Marks & Spencer , enhancing customer engagement and driving online sales.
• Led the end-to-end development for high-impact projects such as Toyota , Lexus , and other Al-Futtaim internal systems—owning architecture, coding, testing, and deployment.

Major Accomplishments:

• Led the WebSphere Portal migration from version 6 to 8, ensuring complete data consistency and minimal downtime during cutover.
• Developed and deployed automated deployment solutions , accelerating delivery cycles and reducing manual intervention.
• Created a custom Maven plugin for deployment automation, which became a widely adopted reusable component across Cognizant teams, significantly improving DevOps efficiency.