Thierno Ibrahima LO

Customer support overwhelmed with repetitive queries, slow response times

Developed intelligent chatbot integrating RAG (Retrieval-Augmented Generation) system with LangChain for context-aware responses. Trained on company knowledge base.

Automated 60% of common queries, reduced support ticket volume significantly

Outdated interface causing poor user experience and low engagement

Designed and implemented modern, responsive interfaces using React, Next.js, and Tailwind CSS with focus on accessibility and performance.

40% improvement in user satisfaction scores, faster page loads

Critical backend code difficult to maintain, no audit trail, database couldn't reproduce past states

Refactored Python backend (Tornado/FastAPI) for better maintainability. Progressively migrated PostgreSQL from traditional relational model to event logging system - zero data loss, zero downtime.

System can now reproduce any past state. Reduced technical debt significantly. Debugging became dramatically easier.

Support team constantly needed developer intervention for routine tasks

Created suite of back-office applications with simple UIs empowering support team to handle tasks independently.

Support team autonomy increased, dev interruptions reduced by 70%

Docker builds taking ages, slow development cycles, noisy logs drowning critical alerts

Optimized Docker build times by orders of magnitude. Implemented one-click CI/CD pipeline with GitHub Actions. Introduced build versioning, unified logging with proper levels and central aggregation.

30% reduction in deployment cycles. 60% reduction in operational noise. Team sees actionable alerts instead of spam.

Inconsistent testing practices, knowledge silos in the team

Started weekly 'Software Testers Club' newsletter with illustrated guides. Facilitated individual kata sessions. Developed reliable test suites following test pyramid pattern.

85% code coverage with TDD. Team-wide improvement in testing skills and practices.

Monolithic payment system couldn't scale for international expansion, deployments took 45+ minutes

Broke monolith into 8 properly-bounded microservices communicating via gRPC. Set up Kubernetes orchestration with independent scaling. Implemented comprehensive monitoring with Prometheus and Grafana.

Deploy time: 45 min → 3 min. System now handles 10x traffic. Expanded to 5 new markets. Team knows what's broken before customers do.

Legacy data processing couldn't handle growing transaction volumes, high latency

Implemented Apache Kafka for async event processing. Optimized Cassandra queries from 10+ seconds to <100ms through strategic indexing.

35% latency reduction. 50% improvement in query performance. System ready for scale.

Law firm drowning in paper documents, manual processing, no efficient case tracking

Designed and developed full-stack platform: RESTful APIs with FastAPI/Node.js deployed on GCP. Integrated Google Cloud Vision OCR for scanned document text extraction. Built tools to extract highlights, annotations, and split PDFs by bookmarks. Modern React frontend for streamlined UX. Celery/Redis for async processing of large document batches.

Paperless workflow achieved. Hours saved per case. Lawyers can focus on legal work instead of document wrangling.

Developers wasting hours manually configuring deployments for each new app

Built CLI tool in Golang that auto-generates deployment configs. Implemented gRPC endpoints for service communication. Created Azure deployment scripts with Terraform (IaC). Contributed to extensible plugin system.

Deployment setup: hours → minutes. Team focuses on features, not infrastructure.

Manual deployments taking days, configuration drift everywhere

Deployed FastAPI applications with complete CI/CD pipeline to Azure App Service. Configured Azure DevOps for automated testing and deployment. Implemented Infrastructure as Code with Terraform for Azure Container Registry and AKS clusters.

Deploy time: days → 15 minutes. Configuration drift eliminated. Repeatable, reliable deployments.