Designed and delivered a full-stack MLOps platform for orchestrating large-scale AI model training experiments, targeting research teams running distributed GPU workloads.

• Experiment Orchestration Engine: Architected a distributed training job scheduler supporting multi-node GPU cluster allocation (up to 64× H100-SXM), with real-time status tracking, priority queuing (low/normal/high/critical), and hyperparameter versioning - reducing experiment turnaround time by eliminating manual job handoff.
• Model Registry & Lifecycle Management: Built a multi-stage model promotion pipeline (Draft - Validated - Staging - Production - Archived) with approval gates, artifact URL tracking, and dataset version linkage, enabling full ML lineage traceability.
• Dataset Versioning & Governance: Implemented a dataset catalog supporting multi-format ingestion (Parquet, JSONL, TFRecord, HDF5) with schema validation, row-count tracking, and deprecation workflows - directly addressing GDPR/PII data governance requirements.
• Observability & Cluster Monitoring: Delivered a real-time monitoring dashboard surfacing GPU/CPU utilization, memory pressure, and cluster-wide health using time-series charts and per-node metrics - applying the "Three Pillars" observability model (metrics, logs, tracing) consistent with SRE best practices.
• CI/CD Pipeline Visualization: Engineered a pipeline management UI for CI, CD, training, and evaluation runs, with Git branch/commit traceability, trigger type tracking, and stage-by-stage execution visualization - integrated with GitLab CI/CD webhooks.
• GPU Quota & Cost Management: Built a team-level GPU quota management system with utilization alerting (threshold-based), priority-aware scheduling visualization, and allocation charts - enabling FinOps-level resource governance across research teams.
• Secure, Cloud-Native Architecture: Applied hardened Docker multi-stage builds, non-root container users, JWT-based API security, and role-based access control (RBAC) throughout the stack, consistent with DevSecOps standards (SAST/OWASP compliance).

Key Technologies:

• React + TypeScript: SPA frontend with OnPush-equivalent optimization patterns, real-time WebSocket subscriptions for live job metrics
• Spring Boot (Java 17): RESTful API layer with idempotency keys, Resilience4j circuit breakers, and distributed tracing (correlation IDs)
• PostgreSQL: Schema-optimized for time-series experiment metrics and lineage DAGs; Hibernate batch operations for high-volume telemetry ingestion
• Kubernetes + Helm: Containerized microservices deployment with horizontal pod autoscaling for training workers
• Prometheus + Grafana: Cluster metrics pipeline with custom alerting rules and SLO dashboards
• GitLab CI/CD: Automated security scanning (SAST, DAST, SCA), secret detection, and quality gates on every push

• Undertook a structured, hands-on upskilling phase to strengthen modern full-stack development and quality engineering skills.
• Completed end-to-end development and deployment of a full-stack, AI-powered Software-as-a-Service (SaaS) application. The platform enables users to interact conversationally with their PDF documents, leveraging advanced Retrieval-Augmented Generation (RAG) to provide accurate, context-aware answers sourced directly from the user-uploaded documents, eliminating AI hallucinations.

Technical Implementation & Key Responsibilities

• Full-Stack System Architecture
  • Architected a scalable, full-stack system using a modern tech stack including Next.js 13+ (App Router) with TypeScript for the frontend and API layer, Tailwind CSS and shadcn/ui for the responsive UI, and Clerk for secure authentication.
• Data & AI Pipeline Engineering
  • Engineered a robust data and AI pipeline by integrating Neon (Serverless PostgreSQL) with Drizzle ORM for type-safe database operations and AWS S3 for secure, scalable file storage.
  • Designed and implemented the core AI functionality using the OpenAI API and LangChain framework. Developed a custom RAG system where PDFs are processed, chunked, and their embeddings stored in Pinecone (vector database), enabling the AI to retrieve relevant document context for each user query.
• Real-Time User Experience
  • Integrated real-time, streaming chat interfaces using the Vercel AI SDK, delivering a seamless, ChatGPT-like user experience where AI responses are streamed directly to the client.
• Monetization & Business Logic
  • Implemented monetization and subscription management by integrating the Stripe API, including checkout flows, webhooks for payment event handling, and a customer portal, enabling recurring revenue.
• Deployment & DevOps
  • Established a full CI/CD pipeline by connecting the GitHub repository to Vercel, enabling automatic builds and deployments on every push to the main branch, ensuring rapid and reliable iteration.

Key Technologies

• Next.js: Served as the full-stack React framework for server-rendered frontend and API routes, chosen for performance and seamless Vercel integration.
• TypeScript: Ensured type safety and improved maintainability throughout the complex codebase.
• OpenAI API: Powered core AI functionality for generating chat responses and creating text embeddings.
• LangChain: Orchestrated the document processing pipeline to load and split PDFs into chunks.
• Pinecone: Acted as the vector database for high-speed similarity search on document embeddings.
• Stripe: Handled all subscription billing and payment processing for the "Pro" tier.
• AWS S3: Utilized for secure and scalable cloud storage of user-uploaded PDF files.
• Neon (PostgreSQL): Provided the serverless, scalable relational database for application data.
• Drizzle ORM: Used as a TypeScript-first ORM for type-safe and efficient database interactions.
• Tailwind CSS: Enabled rapid development of a custom, responsive UI with a utility-first approach.
• Clerk: Implemented for secure and streamlined user authentication and session management.
• Vercel AI SDK: Simplified the implementation of the real-time, streaming chat interface.
• Vercel: Chosen as the deployment platform for its global edge network and native CI/CD.

Ensured the stability and modernization of a mission-critical workforce management platform, working across backend systems, integrations, and DevOps pipelines.

• Systems Stability:
  • Maintained and improved a large-scale Java application in production, resolving incidents and implementing durable fixes.
• Critical Integrations:
  • Built and operated Java integrations with ATOSS Time Control, ensuring reliable data flow via SOAP web services.
• CI/CD & Automation:
  • Designed GitLab CI/CD pipelines, introduced Docker-based builds, and integrated automated testing and SonarQube quality gates.
• Modernization:
  • Supported the transition from monolith to services, prototyping Spring Boot REST APIs for improved scalability.
• Developer Efficiency:
  • Created internal Eclipse plugins to automate workflows and enforce standards.
• Impact:
  • ?Strengthened production reliability, deployment safety, and operational foundations in a strict enterprise environment.

Drove full-stack development and system optimization for healthcare software solutions, bridging backend APIs, web/mobile applications, and DevOps. Translated deep technical insights from support into robust engineering improvements, enhancing performance, scalability, and deployment reliability.

• Engineered Scalable Backend Systems:
  • Developed and optimized secure RESTful APIs using Java/Spring Boot, integrating Hibernate and Ehcache to enhance performance and reduce database load. 
  • Designed efficient PostgreSQL schemas to support complex data workflows.
• Delivered Integrated Full-Stack Features:
  • Built responsive web interfaces with Angular and Angular Material for internal tools.
  • Extended platform functionality by developing companion mobile applications with Flutter, implementing BLE communication and local data persistence.
• Architected & Modernized Deployment Pipelines:
  • Improved system resilience and deployment speed by designing and implementing a microservices architecture. 
  • Automated build, test, and deployment processes using Docker, Jenkins, and Ansible, establishing a reliable CI/CD foundation.
• Solved Critical Integration & Support Challenges:
  • Integrated real-time RFID/BLE asset tracking via custom APIs, enabling seamless data flow between legacy and modern systems. 
  • Leveraged support insights to proactively improve data synchronization, reduce latency, and optimize system-wide performance.
• Championed Quality & Collaboration:
  • Elevated code quality through rigorous peer reviews and the development of comprehensive test cases. 
  • Actively contributed to Agile workflows, ensuring consistent delivery of stable software across multiple product versions.

Engineered and evolved full-stack web applications from a Perl foundation to modern Java/Spring Boot and Angular systems. Drove platform modernization by establishing automated CI/CD pipelines and championing quality-first development practices.

• Built & Modernized Scalable Applications:
  • Spearheaded development using Perl (Mojolicious) and later Java 11/17, Spring Boot, and Angular 13+, optimizing MySQL/PostgreSQL databases for performance and reliability.
• Engineered Secure & High-Performance APIs:
  • Developed RESTful APIs with Spring Boot (JWT Security) and Perl, integrating third-party services. Enhanced backend efficiency using Hibernate ORM and Ehcache caching.
• Developed Engaging User Interfaces:
  • Created responsive, dynamic frontends with Angular (Material, RxJS) and core JavaScript/jQuery/CSS for seamless cross-platform user experiences.
• Automated DevOps & Deployment:
  • Accelerated release cycles by containerizing apps with Docker and implementing CI/CD pipelines with Jenkins/Ansible for automated testing and deployment.
• Championed Code Quality & Reliability:
  • ?Established TDD practices, rigorous testing with JUnit/Mockito/Postman, and Git-based collaboration to ensure system resilience and operational excellence.