# Sundara Raman Ramachandran > LLM Inference/Recommendation Systems @ LinkedIn | Staff Engineer | SGlang Contributor | UT Austin Location: San Francisco Bay Area, United States Profile: https://flows.cv/sundara ๐Ÿ“ฉ Feel free to reach out to me at ๐ฌ๐ฎ๐ง๐๐š๐ซ๐Ÿ๐Ÿ’๐Ÿ๐Ÿ—๐Ÿ“@๐ ๐ฆ๐š๐ข๐ฅ.๐œ๐จ๐ฆ I regularly speak, write, and collaborate on ๐‹๐‹๐Œ ๐ข๐ง๐Ÿ๐ž๐ซ๐ž๐ง๐œ๐ž, ๐ซ๐ž๐œ๐จ๐ฆ๐ฆ๐ž๐ง๐๐ž๐ซ ๐ฌ๐ฒ๐ฌ๐ญ๐ž๐ฆ๐ฌ, and ๐€๐ˆ ๐ข๐ง๐Ÿ๐ซ๐š๐ฌ๐ญ๐ซ๐ฎ๐œ๐ญ๐ฎ๐ซ๐ž. I build ๐ก๐ข๐ ๐ก-๐ฉ๐ž๐ซ๐Ÿ๐จ๐ซ๐ฆ๐š๐ง๐œ๐ž ๐‹๐‹๐Œ ๐ข๐ง๐Ÿ๐ž๐ซ๐ž๐ง๐œ๐ž ๐ฌ๐ฒ๐ฌ๐ญ๐ž๐ฆ๐ฌ for ๐ซ๐ž๐š๐ฅ-๐ญ๐ข๐ฆ๐ž ๐ซ๐š๐ง๐ค๐ข๐ง๐  ๐š๐ง๐ ๐ซ๐ž๐œ๐จ๐ฆ๐ฆ๐ž๐ง๐๐š๐ญ๐ข๐จ๐ง ๐ฐ๐จ๐ซ๐ค๐ฅ๐จ๐š๐๐ฌ. Iโ€™m a ๐’๐ญ๐š๐Ÿ๐Ÿ ๐„๐ง๐ ๐ข๐ง๐ž๐ž๐ซ on ๐‹๐ข๐ง๐ค๐ž๐๐ˆ๐งโ€™๐ฌ ๐‹๐‹๐Œ ๐ˆ๐ง๐Ÿ๐ž๐ซ๐ž๐ง๐œ๐ž ๐ญ๐ž๐š๐ฆ, focused on serving large language models in production with low latency, high throughput, and efficient GPU utilization. My work spans the LLM serving stackโ€”including tokenization, batching, scheduling, prefill-only scoring, and system-level performance optimization at scale. Iโ€™m an active ๐จ๐ฉ๐ž๐ง-๐ฌ๐จ๐ฎ๐ซ๐œ๐ž ๐œ๐จ๐ง๐ญ๐ซ๐ข๐›๐ฎ๐ญ๐จ๐ซ to ๐’๐†๐‹๐š๐ง๐ , upstreaming production-driven performance improvements and APIs that turn high-QPS constraints into reusable infrastructure. Previously, I worked on ๐€๐ณ๐ฎ๐ซ๐ž Identity & Authorization and ๐Œ๐ข๐œ๐ซ๐จ๐ฌ๐จ๐Ÿ๐ญ ๐Ž๐Ÿ๐Ÿ๐ข๐œ๐ž. I hold a ๐Œ๐š๐ฌ๐ญ๐ž๐ซโ€™๐ฌ ๐๐ž๐ ๐ซ๐ž๐ž from ๐“๐ก๐ž ๐”๐ง๐ข๐ฏ๐ž๐ซ๐ฌ๐ข๐ญ๐ฒ ๐จ๐Ÿ ๐“๐ž๐ฑ๐š๐ฌ ๐š๐ญ ๐€๐ฎ๐ฌ๐ญ๐ข๐ง. ## Work Experience ### Contributor @ SGLang Jan 2025 โ€“ Present Contributed performance and API improvements across tokenization, batching, scoring, and benchmarking in the SGLang LLM serving framework. 1. Tokenizer: Batch tokenization and dynamic batching https://github.com/sgl-project/sglang/pull/5141 https://github.com/sgl-project/sglang/pull/9382 2. ZMQ Pipeline: Batched send from Tokenizer Manager https://github.com/sgl-project/sglang/pull/9436 3. Generative Score API: Removed decode path, optimized prefill-only scoring, added multi-item scoring with custom attention masks https://github.com/sgl-project/sglang/pull/8840 https://github.com/sgl-project/sglang/pull/9748 https://github.com/sgl-project/sglang/pull/10979 4. Benchmarking: Prefill-only benchmark scripts https://github.com/sgl-project/sglang/pull/10240 ### Staff Software Engineer @ LinkedIn Jan 2025 โ€“ Present | Mountain View, CA LLM Serving Infra (on-prem and cloud) @ LinkedIn ### Senior Software Engineer @ LinkedIn Jan 2022 โ€“ Jan 2025 | Sunnyvale, California, United States LLM Serving Infra (on-prem and cloud) @ LinkedIn ### Software Engineer 2 @ Microsoft Jan 2021 โ€“ Jan 2022 | Redmond, Washington, United States Member of Azure Identity and Access Management team. Working on RBAC Authorization service. ### Graduate Teaching Assistant @ The University of Texas at Austin Jan 2019 โ€“ Jan 2020 | Austin, Texas Area Graduate Teaching Assistant at the Computer Science department. ### Software Engineer Intern @ Google Jan 2020 โ€“ Jan 2020 | Mountain View, California, United States ### Software Engineer 2 @ Microsoft Jan 2018 โ€“ Jan 2019 | Hyderabad Area, India ### Software Engineer @ Microsoft Jan 2016 โ€“ Jan 2018 | Hyderabad Area, India Part of the feature team which designed and developed "Data Visualizer" & "Microsoft Flow Designer" Microsoft Visio Desktop Application. Team size: Developers(15) + Program Manager(1) + Manager(1). ### Mathematics Teacher @ Arul Institute - India Jan 2012 โ€“ Jan 2014 | Chennai Area, India Arul Institute is a private tuition center, located in Chennai, Tamil Nadu, India, which provides tuition, mentoring and guidance for Higher Secondary and Secondary school students. 1. Worked as a Mathematics teacher, mentor & counselor for 10th standard matriculation students. 2. Assisted 10 of my students to achieve 100 percent marks in their 10th standard board exams. ## Education ### MS in Computer Science The University of Texas at Austin ### Undergraduate in Computer Science and Engineering College of Engineering, Guindy ### Higher Secondary in Computer Science Smt. Ramkuwar Devi Fomra Vivekanda Vidyalaya, Chromepet, Chennai. ## Contact & Social - LinkedIn: https://linkedin.com/in/sundara-raman-ramachandran --- Source: https://flows.cv/sundara JSON Resume: https://flows.cv/sundara/resume.json Last updated: 2026-04-01