MUHAMMAD USMAN HAIDER

Software development company building scalable web applications and AI-powered solutions. • CloserCoach (AI sales training platform): Architect and develop the backend infrastructure using Node.js, Express.js, and Firebase, supporting an AI-powered sales coaching product with daily active enterprise users. • Integrate VAPI for real-time voice call processing and Langfuse for prompt management and evaluation, enabling AI-driven analysis of sales conversations with sub-2-second feedback latency. • Design and ship the admin dashboard with real-time analytics, user management, call monitoring, and billing views using Firebase Realtime Database. • Build a webhook ingestion layer for VAPI callbacks with retry logic and idempotency handling, processing voice transcriptions and AI-generated feedback reliably. • Develop an AI text-enhancement tool using OpenAI GPT models to improve grammar, tone, and clarity of user-submitted scripts in real time. • Build a property booking platform with availability management, Stripe payment integration, and automated host-guest messaging. • Implement a serverless backend using the SST framework with AWS Lambda, API Gateway, and DynamoDB to reduce infrastructure costs and scale independently. • Build an intelligent form auto-fill bot using NLP and pattern matching to extract structured data from PDFs, emails, and spreadsheets.

• Built a full-stack RAG system using Next.js, LangChain, and Pinecone for intelligent Q&A across personal document libraries with semantic search. • Implemented a document ingestion pipeline supporting PDF, DOCX, TXT, and Markdown with chunking strategies and OpenAI text-embedding-3-small for vector embeddings. • Architectural semantic retrieval over Pinecone with GPT-4-powered answer generation and source citations to keep responses grounded. • Developed a Next.js frontend with drag-and-drop file upload, streaming query responses, and conversation history.

• Built a full-stack AI optimization platform (founded by an ex-Amazon Head of SEO) tracking brand visibility and citations across ChatGPT, Claude, Gemini, and Perplexity. • Developed the backend in NestJS with PostgreSQL and a Bull queue system for brand tracking, prompt testing, and async job orchestration over multi-LLM workloads. • Architected the marketing site and analytics dashboard in Next.js with real-time data visualization for monitoring brand perception across LLM platforms. • Integrated Sentry for error tracking, Datadog for APM and infrastructure monitoring, Intercom for customer support, and Clerk for authentication. • Integrated OpenAI GPT-4, Claude, Gemini, and Perplexity APIs to deliver automated brand visibility insights, competitive benchmarking, and reputation alerts.

IT solutions company focused on healthcare automation products. • Developed and maintained 3 microservices for fax processing and appointment management using NestJS and TypeScript, as part of a 20-person distributed team across the US, India, and Pakistan. • Implemented event-driven communication between microservices using Apache Kafka for real-time healthcare document processing, significantly reducing manual triage time. • Designed a polyglot persistence layer using MongoDB for document storage, PostgreSQL for relational data, and Redis for caching, improving query latency on hot paths. • Built RESTful APIs and integrated with major EMR systems (Oscar, Accuro, Athena Health) to enable healthcare data exchange and automated fax triage for clinical workflows. • Collaborated in Agile sprints with cross-functional teams across 3 time zones, consistently shipping features on time while maintaining high API uptime for enterprise healthcare clients.

AWS Certified Machine Learning Engineer - Associate

AWS