Headline

Software Engineer @ Adobe Premiere Pro | IIIT-H Alum | C++ · JS · GenAI · High-Performance Video Systems

About

I build AI-powered video features for Adobe Premiere Pro — used by millions of professional editors and filmmakers worldwide. Over 3+ years at Adobe, I've shipped production features that moved real numbers: · Reduced manual rotoscoping by ~90% via GPU-accelerated AI Object Masking · Contributed to a 2% MAU increase across Premiere Pro through multi-language Captions Translation · Cut manual audio tagging by ~90% with an Automatic Audio Category Tagger at 100% telemetry coverage My work spans low-level C++ systems, on-device AI inference, full-stack JavaScript/React integrations, and cross-functional AI feature delivery from scoping to production. Now exploring Senior SDE roles in AI-powered product engineering, media tech, or developer tools where I can own systems end-to-end and lead high-impact teams.

Location

Bengaluru

AI stack

antigravity

codex

cursor

Work experience

CS1 at Adobe, Jan 2024 - Present · 2 yrs 8 mos, Bengaluru, Karnataka, India, • Engineered AI Object Masking in C++, reducing manual rotoscoping by ~90% while preserving real-time playback. • Led the end-to-end Captions Translation integration using C++, JavaScript, and React, driving a 2% growth in MAU. • Drove on-schedule delivery of AI features across multiple Premiere Pro releases, aligning product, research, and platform teams while mentoring junior engineers.

Member of Technical Staff at Adobe, Jul 2022 - Present · 4 yrs 2 mos, Reduced manual video rotoscoping by ~90% - measured by QA benchmark tests across 500+ clips - by engineering AI Object Masking in C++ with GPU-accelerated on-device inference with zero regression in real-time playback performance. Increased MAU by 2% by leading end-to-end Captions Translation (C++, JavaScript, React), designing OOP abstractions to fully decouple translation logic from the UI layer and shipping across 10+ locales. Cut manual audio-tagging effort by 90% and improved audio organisation accuracy by 30% — validated via A/B telemetry at 100% instrumentation coverage — by building the Automatic Audio Category Tagger with concurrency-optimised inference and enforcing best-practice code reviews across the team.

MTS1 at Adobe, Jul 2022 - Jan 2024 · 1 yr 7 mos, Bengaluru, Karnataka, India, • Developed Premiere Pro's Automatic Audio Category Tagger, significantly reducing manual audio tagging effort by ~90%. • Implemented concurrency optimizations and Jenkins CI/CD pipelines to enhance audio organization by 30%. • Achieved full automation of a previously manual editorial workflow, streamlining processes for better efficiency.

Research Intern at Adobe Research Labs, May 2021 - Jul 2021 · 3 mos, Achieved 2% improvement in entity extraction F1-score on low-resource domains — validated across 3+ domain-specific benchmarks — by designing and running experiments on one-shot learning techniques (prototypical networks, in-context prompting) to adapt NLP models with minimal labelled data, directly reducing annotation effort by ~40%. Systematically compared few-shot vs. fine-tuned baselines through controlled ablation studies, identifying one-shot transfer as the optimal trade-off between annotation cost and accuracy; findings informed domain-adaptive interface design adopted by downstream research and engineering teams.

Research Intern at ISRO, RRSC-West, Jun 2019 - Jul 2019 · 2 mos, Reduced visitor check-in time significantly and handled 100+ daily visitors at peak load — verified through stakeholder UAT — by designing and implementing a concurrent, multilingual Visitor Management System (Python, Django, SQL, JavaScript) with role-based authentication, scheduling, and pass generation across 3+ languages.

Education

Master of Technology, Computer Science, International Institute of Information Technology Hyderabad (IIITH)

BE - Bachelor of Engineering, Computer Science, M.B.M. Engineering College, Jodhpur