Ayaan Khan

Aligarh, IN

• Designed and implemented a highly scalable, event-driven multi-stage execution pipeline using AWS ECS, ECR, and AWS Device Farm. The architecture successfully normalizes heterogeneous CI/CD triggers into a unified payload contract. To ensure high availability and resource isolation, I integrated queue-based routing, priority task scheduling, and dead-letter queue (DLQ) retry mechanisms, protecting against execution bottlenecks and runtime infrastructure failures.

• Built an autonomous end-to-end AI testing agent using LangGraph.js, Playwright, and multi-agent coordination models to automatically traverse web applications and generate test flows. To overcome the high token costs and latency of sending massive raw DOM trees to LLMs, I engineered a custom DOM-pruning pre-processor. The script filters out non-viewport components, non-interactable tags, and background layers, reducing the context payload by 85% while boosting reasoning accuracy.

• Architected a hybrid self-healing engine for web automation selectors to dynamically repair broken testing scripts during execution. The system features a deterministic heuristic layer calculating Levenshtein distance, cosine similarity of element attributes, and spatial layout proximity. For complex UI layout shifts where heuristics fail, the engine triggers selective, low-latency LLM fallbacks, maintaining execution reliability while minimizing API consumption costs.

• Refactored a monolithic, web-only test runner into a unified, cross-platform executor supporting Web, Mobile, and API test suites. By utilizing registry, Inversion of Control (IoC), and strategy design patterns, the system cleanly decouples execution logic and shares runtime context across mixed-platform workflows. I also developed independent, asynchronous event listeners to handle execution log streaming and real-time database persistence.

• Engineered a custom nested evaluation parser and interpolation engine from scratch to power complex, data-driven automation steps. I designed recursive evaluation algorithms capable of resolving system variables, user-defined generator functions, and external tabular datasets (CSV/Excel) on the fly during test execution. This allows users to build highly flexible workflows that adapt dynamically to varying runtime inputs.

• Developed a role-based Identity & Access Management (IAM) system coupled with a resilient, real-time usage-based billing and metering pipeline. By integrating with Stripe, the pipeline automatically tracks granular API consumption, handles subscription tier limits, and flags overages. The background pipeline was built using resilient event queues to guarantee that quota enforcement occurs in real-time without introducing latency to user requests.

• Architected the release pipeline by building automated, multi-platform CI/CD workflows using GitHub Actions to package, code-sign, and publish desktop applications across macOS, Windows, and Linux. I resolved complex native library compilation issues and integrated client-side auto-updaters with the release pipeline. This automation slashed release packaging times from 8 hours of manual overhead to an automated 1.5-hour process.

Aligarh, IN

• Spearheaded the transition to single-tenant enterprise delivery by developing automated cloud infrastructure orchestration scripts. Using AWS SDKs and custom provisioning templates, I automated the single-click setup of complete client ecosystems consisting of EC2 instances, MongoDB, Redis clusters, Lambda functions, S3 buckets, SQS queues, and custom Nginx configurations. This reduced manual onboarding time for premium enterprise clients to just a few minutes.

• Refactored the core lead-ingestion APIs to process high-throughput, unstructured payloads originating from diverse third-party integrations. I redesigned the database ingestion routines to gracefully handle write spikes, eliminating processing failures and improving data consistency. Additionally, I built an interactive, multi-dimensional analytics dashboard allowing clients to track real-time distribution metrics and operational bottlenecks.

• Implemented a multi-layer caching strategy to minimize database latency and API response times across the TeleCRM ecosystem. The architecture pairs client-side offline storage (IndexedDB managed via Dexie) with server-side caching using Redis. When integrated with AWS CloudFront edge caching, the system realized a reduction in overall page load latency of up to 70%, yielding a highly responsive user experience.

• Designed and developed an extensible integration and outbound webhook module from scratch to support custom workflows. The module enables users to trigger RESTful requests (GET, POST, PUT, DELETE) on external servers based on conditional events occurring inside their CRM. I implemented a robust scheduler and retry worker queue to guarantee payload delivery even if target third-party servers experience transient downtime.

Bristol, England

• Worked within the Amazon IMDb data coverage team on large-scale data integrity.

Remote

• Engineered a serverless full-stack web application comparing nationwide utility prices, utilizing AWS (API Gateway, DynamoDB, Cognito) for secure backend APIs and implementing graph algorithms to dynamically render JSON-driven forms.

Developed a monolithic x86 (IA-32) operating system kernel, implementing low-level systems programming concepts including interrupt handling (IDT/PIC), virtual memory management, multitasking, system calls, DMA, and device drivers for PCI, keyboard, and VGA display.

Patched framework-level UI and Material Design bugs, notably implementing optimizations for Material Design swatch color comparisons (MaterialColor Swatch Map) and resolving widget-rendering edge cases.

Contributed low-level computational geometry and spatial indexing optimizations, resolving boundary calculations and polygon intersection edge cases in the C++ Boost library.

Patched core routing mechanisms, template compilation rules, and server thread lifecycle behaviors to prevent connection leaks and uncaught exceptions in high-throughput C++ microservices.

Developed and optimized fundamental computer science algorithms in C++, Java, and JavaScript; mentored first-time contributors and reviewed over 100 pull requests for code quality, optimization, and style correctness.