Headline

Lead Software Engineer @ Adobe | AI Evaluation & LLM Benchmarking | AutoEval Frameworks | Python Automation | CI/CD Infrastructure | Distributed Systems

About

I am a Lead Software Engineer at Adobe with 6+ years of experience building large-scale automation systems, CI/CD infrastructure, and AI evaluation frameworks. My recent work focuses on developing evaluation systems for AI-powered research agents. I designed and implemented a Python-based AutoEval framework and scalable LLM-judge pipelines that benchmark AI models across reasoning, synthesis, and source attribution tasks. These systems use prompt-engineered evaluation workflows and multi-judge consensus to automate qualitative assessment of AI-generated responses. Previously, I worked extensively on CI/CD infrastructure supporting Adobe Acrobat PDF processing and tagging pipelines across Windows, macOS, and Linux environments. I also architected containerized stability testing frameworks and built internal tools that improved automation reliability and debugging efficiency across distributed build environments. I enjoy building engineering systems that improve developer productivity, automate complex workflows, and enable reliable evaluation of modern AI systems. Key areas of interest • AI Evaluation & LLM Benchmarking • Automation Infrastructure • Python Engineering • CI/CD Systems • Distributed Systems • Developer Productivity Tooling I’m always interested in conversations around AI systems, automation infrastructure, and large-scale engineering challenges.

Location

Noida

Work experience

Lead Software Engineer at Adobe, Feb 2025 - Present · 1 yr 5 mos, Noida, Uttar Pradesh, India, • Designed and implemented a Python-based AutoEval framework for Deep Research AI agents, defining structured evaluation metrics across research comprehension, multi-hop reasoning, insight synthesis, and document/web attribution to benchmark performance across LLM systems. • Built scalable LLM-judge evaluation pipelines using prompt-engineered GPT and Claude models, enabling multi-judge consensus and dataset-driven evaluation to automate qualitative assessment of AI-generated research outputs. • Collaborated with research teams to standardize evaluation workflows for AI research agents, enabling consistent benchmarking and stability comparisons across model iterations. • Developed heuristic-based algorithms to extract slide templates, layouts, and styling metadata from PowerPoint files, enabling automated and scalable presentation generation while maintaining design consistency.

Software Engineer 2 at Adobe, Jul 2021 - Feb 2025 · 3 yrs 8 mos, Noida, Uttar Pradesh, India, • Managed and optimized CI/CD pipelines for Adobe Acrobat PDF tagging components, integrating automated unit testing, objective quality metrics, and structured error reporting across Windows, macOS, and Linux environments. • Architected and deployed a containerized Stability Testing Framework using Docker, improving automation reliability and enabling advanced analytics and reporting through Tableau dashboards. • Improved build reliability by rapidly diagnosing and resolving CI pipeline failures across distributed build environments.

Software Engineer at Adobe, Sep 2019 - Jul 2021 · 1 yr 11 mos, On-site, • Developed an HTML-based visual diff tool using Python, JavaScript, and HTML to compare Liquid Mode rendering outputs across PDF versions and highlight structural differences. • Led automation of quality and qualifier pipelines on real Android devices, reducing manual configuration dependency by 50% and improving automated test execution reliability. • Built internal automation tools to support large-scale PDF rendering validation workflows.

Education

Bachelor of Engineering - BE, Computer Science, CHANDIGARH UNIVERSITY