Mohammad Usman

Senior Software Engineer specializing in AI-powered platforms and high-performance cloud infrastructure.

Based in: Lahore, PakistanCurrently: Senior Software Engineer, Beyond ONEExperience: 6 years

Senior Software Engineer with over 6 years of experience architecting scalable AI/LLM applications, RAG pipelines, and full-stack systems across AWS and GCP. Proven track record in optimizing platform performance, reducing API latency by up to 92%, and leading engineering teams to deliver high-impact digital products.

Experience

Senior Software Engineer

Beyond ONELahore, PakistanworkJun 2025 – Present

Summary

Built and scaled digital telecom products across multiple brands and markets, working across full-stack engineering, platform architecture, DevOps, observability, and developer experience.

What I did

Led development of FRiENDi Mobile KSA from early scaffolding to a customer-facing platform, delivering core purchasing, recharge, support, multilingual, and SEO experiences while building session, eligibility, and payment flows.
Engineered the FRiENDi self-activation funnel end-to-end, from SIM validation and eligibility through identity verification, plan selection, number selection, and payment; expanded E2E automation and reduced regression issues by ~35%.
Contributed to Travel eSIM, building Next.js product experiences and owning analytics, security, CI/CD, deployment, and production readiness, contributing to ~40% fewer release defects.
Architected shared foundations for a multi-brand web monorepo, establishing reusable application, API, design-system, CMS, i18n, and E2E testing patterns, improving development velocity by ~30% across brands.
Built CI/CD and engineering automation with standardized pipelines, performance monitoring, automated releases, Docker builds, and deployment workflows, reducing manual release effort by ~50%.
Strengthened production reliability through Sentry, Dynatrace, Amplitude, feature flags, and monitoring, while driving AI-powered engineering automation including Claude-based PR review, automated Sentry issue resolution, and Kubernetes deployments.

Results

Reduced regression issues by ~35% through E2E automation.
Contributed to ~40% fewer release defects for Travel eSIM.
Improved development velocity by ~30% across brands via monorepo architecture.
Reduced manual release effort by ~50% through CI/CD automation.

Next.jsCI/CDDockerKubernetesSentryDynatraceAmplitudeClaudeMonorepoCMSi18n

Senior Software Engineer - AI and Growth

Makula TechnologyLahore, PakistanworkJun 2024 – Jun 2025

Summary

Architected an AI-powered semantic search platform using vector embeddings and RAG pipelines, improving search accuracy by 40% and reducing response latency by 55%. Developed Python-based backend services and REST APIs for AI applications, integrating vector databases and document processing pipelines. Designed an AI co-pilot assistant using RAG architecture, reducing customer support response time by 60%. Built a scalable OCR document processing pipeline using Azure Document Intelligence and AWS SageMaker, processing 10K+ documents monthly. Deployed containerized AI services using Docker and Kubernetes, enabling scalable infrastructure for high-volume AI workloads. Refactored monolithic architecture into modular services using Yarn Workspaces, improving engineering productivity by 30%. Optimized analytics APIs with MongoDB aggregation pipelines, reducing data retrieval time by 60%.

What I did

Optimized document chunking by transitioning from page-by-page to paragraph-by-paragraph processing to improve search accuracy.

Results

Improved search accuracy by 40%.
Reduced response latency by 55%.
Reduced customer support response time by 60%.
Processed 10K+ documents monthly via OCR pipeline.
Improved engineering productivity by 30% through modular refactoring.
Reduced data retrieval time by 60% via MongoDB optimization.

PythonREST APIsVector EmbeddingsRAGAzure Document IntelligenceAWS SageMakerDockerKubernetesYarn WorkspacesMongoDB

Tech Lead - Learner Experience, AI & Performance

EducativeLahore, PakistanworkMar 2023 – Jun 2024

Summary

Led the engineering team delivering B2C and B2B learning platform features, focusing on performance optimization and AI-powered functionality. Designed and launched an AI Mock Interview platform using LLM integrations, increasing lead generation by 14%. Implemented AI-powered semantic search using Elasticsearch and LLM pipelines. Led platform-wide performance optimization initiative using Redis caching and Cloudflare edge caching, reducing API latency from 3 seconds to 250ms (92% improvement). Built scalable backend services integrating search analytics and automated feedback pipelines, improving subscription renewals by 20%. Coordinated cross-functional teams for the DevPath product launch, ensuring successful delivery and platform scalability.

What I did

Implemented LLM streaming and WebSockets to minimize perceived latency for live user experiences.
Utilized Markdown to display formatted LLM output segments in real-time.
Optimized LLMs for the Mock Interview platform using system and output prompting to control model behavior.
Owned the full-text search feature end-to-end from initial build to delivery.

Results