Mukul Nag

About Mukul Nag

Software Engineer with 3+ years spanning firmware validation and AI-driven automation. At Qualcomm, I\'m the sole automation contributor for App Compatibility testing on X Elite ARM-based laptops, having built 4 enterprise tools deployed organization-wide. Previously at AMD, I validated BIOS/UEFI firmware for EPYC server platforms, earning 5 Spotlight Awards.QUALCOMM - AUTOMATION & AI INTEGRATION• Sole Automation Contributor• Built 4 production-ready tools deployed organization-wide• First to integrate LLM-powered analysis into testing workflows• Architected complex multi-agent LLM systems through advanced prompt engineeringTOOLS DEVELOPED1. AI-Driven Test Automation Framework CrewAI multi-agent| faster test creation | 3x coverage | 85-90% self-healing | Natural language → tests | 15+ AI models2. QCompare - AI File Comparison Replaces Beyond Compare licensing | 75% faster reviews (30min→5min)| 90MB standalone | 50+ editor features | 15+ AI models | Offline-capable3. Trending Apps Framework MS Store automation | ARM64 architecture detection | 60% time savings4. AppQompat - Performance Testing Sub-millisecond accuracy launch latency (95%)| Multi-architecture (x86/x64/ARM64)| 70% CPU reduction | 5k+ apps automated testingAMD - FIRMWARE VALIDATION (3 Years)Pre-Silicon BIOS/UEFI Validation for EPYC Turin & Venice Servers• Developed 150+ test cases; automated 120+ reducing manual testing by 40%• Earned 5 Spotlight Awards for bug-free firmware delivery and validation excellence• Validated critical modules: Memory Controller, PCIe, CXL, RAS, Power Management, Interrupt Handling, Platform Security, Bootflow, Caliptra• Architected Python/Bash automation framework integrated with Jenkins CI/CD• Collaborated with Firmware Architects on CPU/GPU microarchitecture (Northbridge, Southbridge, NUMA, I/O, Registers)• Contributed to open-source Platform Security (Caliptra)TECHNICAL EXPERTISEAI/ML: Claude Gemini GPT, CrewAI, LiteLLM, Prompt Engineering, Multi-Agent SystemsAutomation: PyAutoGUI, PyWinAuto, UI Automation, Test Framework DesignFirmware: BIOS/UEFI, Platform Security, Bootflow, RAS, Memory Controller, Power Management, PCIe, CXLLanguages: Python, Bash, PowerShell,Tkinter, Jenkins, Selenium, GitSpecialization: ARM64 Windows, Performance Testing, Enterprise Tool DevelopmentIm passionate about building AI-driven automation solutions, designing robust architectures & delivering measurable business impact through innovative technical solutions.

Experience

1. Engineer

   Qualcomm

   May 2025 — Present · Bengaluru, IN

   Core Achievement: Sole automation contributor building 4 production-ready tools deployed organization-wide for Windows on ARM validation on Snapdragon X-Elite platforms.AI/ML Integration & Prompt Engineering• Designed multi-agent LLM system (CrewAI, 5 agents) achieving 70% faster test creation and 3x coverage• Reduced LLM costs by 67% through intelligent prompt design and token optimization• Created self-healing system with LLM reasoning (90% recovery rate)• Integrated 15+ AI models (Claude Gemini GPT) via unified LiteLLM interface• Implemented adaptive prompting adjusting dynamically to UI changesTechnical Implementation• Architected 5-agent system: Planner, UI Analyzer, Executor, Validator, Self-Healing• Implemented 3-tier element detection: PyWinAuto→keyboard → vision• Built custom syntax highlighter (50+ languages) with viewport rendering• Designed progressive sleep optimization reducing CPU 70%• Implemented perceptual hashing (ImageHash) for dynamic UI detection• Created multi-architecture detection: PE header parsing + IsWow64Process2 APIUser Experience & Tools• Delivered modular Tkinter GUI for plug-and-play adoption• Built 90MB standalone executable replacing Beyond Compare licensing• Created portable execution with window-relative coordinates• Designed HTML reporting with embedded screenshotsPerformance & Optimization• 75% faster code reviews (30min→5min) through AI-powered comparison• Optimized LLM tokens by 62.5% through context truncation• Multi-level caching (UI, LLM, screenshots) for 40% faster operations• Sub-millisecond measurement with 95% accuracy (perf_counter)• 60% time savings through intelligent failure detectionImpact: 4 production tools | 70% faster creation | 3x coverage | 90% maintenance reduction | 90% self-healing | 75% faster reviews | 60% time savings | 70% CPU reductionTechnologies: Python, CrewAI, LiteLLM, Claude, Gemini, GPT, PyAutoGUI, PyWinAuto, Tkinter, PyInstaller, ImageHash, Windows API, Git

Education

• Trailhead by Salesforce

  Salesforce Developer Catalyst Plus

• Air Force School , kanpur

  10+2, Science, Science

  2017 — 2019

• Air Force School ,kanpur

  High School, Cbse

  2007 — 2017

• GCET

  Bachelor of Technology Honors (B.Tech Honors), Information Technology

  2019 — 2023