Founder
Founder
Beseam · Full-timeBeseam · Full-time
Aug 2025 - Present · 8 mosAug 2025 to Present · 8 mos
Zurich, SwitzerlandZurich, Switzerland
Revenue Intelligence for Commerce

Core loop:
Observe · Reason · Act · Learn

🤝 If you run Shopify stores or manage contact to get the free audit to know what might be leaking your revenue.

📥 If you are an Investor/partner/VC, please drop a message.
Revenue Intelligence for Commerce Core loop: Observe · Reason · Act · Learn 🤝 If you run Shopify stores or manage contact to get the free audit to know what might be leaking your revenue. 📥 If you are an Investor/partner/VC, please drop a message.
Stealth Startup logo
Co-Founder
Co-Founder
Stealth Startup · Full-timeStealth Startup · Full-time
Feb 2025 - Jan 2026 · 1 yrFeb 2025 to Jan 2026 · 1 yr
Stay tuned...
Stay tuned...
Pyxi logo
Founding Technical Lead 
Founding Technical Lead 
Pyxi · Part-timePyxi · Part-time
May 2024 - Jul 2025 · 1 yr 3 mosMay 2024 to Jul 2025 · 1 yr 3 mos
* Oversee the technical development, design infrastructure and system performance.
* Maching algorithm and AI implementation.
* Manage development teams and coordinate cross-functional efforts.
* Drive technical strategy and innovation to support user experience.
* Oversee the technical development, design infrastructure and system performance. * Maching algorithm and AI implementation. * Manage development teams and coordinate cross-functional efforts. * Drive technical strategy and innovation to support user experience.
Co-Founder and CEO
Co-Founder and CEO
Co-Founder and CEO
Fisca AI · Full-timeFisca AI · Full-time
Jun 2024 - Jan 2025 · 8 mosJun 2024 to Jan 2025 · 8 mos
Zurich, SwitzerlandZurich, Switzerland
AI Medical coding and billing co-pilot


Software Engineering
Software Engineering
GoogleGoogle
Oct 2014 - Apr 2024 · 9 yrs 7 mosOct 2014 to Apr 2024 · 9 yrs 7 mos
Zurich, SwitzerlandZurich, Switzerland
Led a subteam within YouTube Data responsible for processing YouTube’s global traffic (world’s #2 largest), building large-scale infrastructure for log processing and annotation.

Architected and led development of a highly scalable metadata–annotation merge system that processes over 10 PiB/day across streaming and batch pipelines, supporting 400M+ QPS lookups for A/B experiments, ML dataset generation, and analytics. Co-developed a high-performance distributed cache using RDMA-style remote memory access, reducing per-lookup latency from 200–500 µs in traditional memcache-based systems to a consistent ~10 µs, enabling ultra–low-latency, high-QPS metadata retrieval.

Designed a cost-efficient mechanism for updating immutable datasets using a differential log-based model, reducing infrastructure costs by ~90%.

Co-developed a streaming system to construct user sessions, cutting data delivery latency from 30+ hours to ~2 hours.

Led implementation of a log-tailing platform enabling real-time and streaming operations at full YouTube scale.

Improved correctness and reliability across pipelines responsible for attributing views to YouTube videos.

Co-developed YouTube’s real-time view-count feature, reducing update latency from 2–4 hours to ~10 seconds.
Led a subteam within YouTube Data responsible for processing YouTube’s global traffic (world’s #2 largest), building large-scale infrastructure for log processing and annotation. Architected and led development of a highly scalable metadata–annotation merge system that processes over 10 PiB/day across streaming and batch pipelines, supporting 400M+ QPS lookups for A/B experiments, ML dataset generation, and analytics. Co-developed a high-performance distributed cache using RDMA-style remote memory access, reducing per-lookup latency from 200–500 µs in traditional memcache-based systems to a consistent ~10 µs, enabling ultra–low-latency, high-QPS metadata retrieval. Designed a cost-efficient mechanism for updating immutable datasets using a differential log-based model, reducing infrastructure costs by ~90%. Co-developed a streaming system to construct user sessions, cutting data delivery latency from 30+ hours to ~2 hours. Led implementation of a log-tailing platform enabling real-time and streaming operations at full YouTube scale. Improved correctness and reliability across pipelines responsible for attributing views to YouTube videos. Co-developed YouTube’s real-time view-count feature, reducing update latency from 2–4 hours to ~10 seconds.


Software Developement Engineer
Software Developement Engineer
AmazonAmazon
Oct 2011 - Sep 2014 · 3 yrsOct 2011 to Sep 2014 · 3 yrs
Greater Hyderabad AreaGreater Hyderabad Area
Amazon Sort center

* Involved in building Next Generation transportation systems. 
* Developed Scalable distributed Systems for manual and automated sortation of packages. 
* Developed various heuristics and services to intelligently handle packages processing, labor management and defect detection.
* User various AWS systems. 
* Reduced cost of sorting packages by 500%.