May 2025 – Present · 10 mos

• Co-founded PipesHub, an open-source infrastructure for building AI agents on enterprise data.
• Led development of modular connectors, Agentic RAG, retrieval pipelines, and Knowledge Graph.
• Enabled streaming answers with pinpointed citations, simplifying infrastructure for developers.
• 2000+ GitHub stars in 4 months, 25+ companies actively using PipesHub community edition, thousands of AI queries served.

Jan 2024 – Mar 2025 · 1 yr 2 mos

• Co-founded an AI-first LegalTech startup focused on transforming Contract Lifecycle Management (CLM) using cutting-edge Generative AI and Agentic Retrieval-Augmented Generation (RAG) technologies.
• Built a proprietary Contract Co-Pilot and smart metadata extraction system, delivering 70%+ efficiency gains and 50% cost savings for legal teams.
• Achieved early commercial success, generating around $100K in revenue within the first year and securing marquee clients including McDonald’s and HCL

Aug 2021 – Nov 2023 · 2 yrs 3 mos

• Spearheaded latency reduction in FPGA orders reconciliation and also helped in significantly improving core trading library latency.
• Contributed to the design and implementation of a next-generation trading platform, enhancing trading capabilities.

Dec 2017 – Aug 2021 · 3 yrs 8 mos · Bengaluru Area, India

• Led engineering teams for two global trading infrastructure projects, including a Low Latency Monitoring Microservices Stack and a JVM Warmup Framework, driving improvements in system performance and stability for real-time trading.
• Core contributor to GSX Internalization & SigmaX Trading Platform, working on low-latency, zero-GC distributed systems for order crossing within Goldman Sachs.
• Architected and led a team of 5 engineers to build the Next-Generation Exchange Simulator, a microservices-based platform in Java and C++ that emulated exchange behavior, accelerating testing and reducing production defects.
• Spearheaded development of ultra-low latency exchange connectivity systems, migrated trading flows to newer, faster stacks, and implemented robust network failure handling including automatic order cancellation on disconnects.
• Designed and implemented an automatic broker selection mechanism with weighted scheduling algorithms for optimized trade execution.

Jun 2016 – Nov 2017 · 1 yr 5 mos · Bangalore

Jan 2015 – May 2016 · 1 yr 4 mos · Bengaluru Area, India

Aug 2013 – Jan 2015 · 1 yr 5 mos · Bangalore

May 2012 – Aug 2012 · 3 mos · Stuttgart Area, Germany

• The L4 microkernel provides realtime guarantees and was designed
• to be a promising alternative to monolithic kernels. However, drivers developed for monolithic kernels have to be modified to be able to work on L4. The main aim of the project was to get X11(X Window System) running on paravirtualised linux with Graphics Acceleration.

Dec 2011 – Feb 2012 · 2 mos · Sydney

• The motivation for Linux as a Bootloader is simple that the boot process of
• Operating Systems is not good as it could be and there is kexec system call and kexec-tools which can make it do better. One has to rewrite drivers or protocols the bootloader incorporates and has to be maintained in parallel with the original source. In solutions like Linux as a Bootloader one could reuse the drivers or protocols written for Linux.The aim of the project was to be able to boot not just Linux but any other Operating System from
• Linux on ARM and x86 architecture,