Personal site of Cyan Subhra Mishra — Performance and Power Engineer at Arm in San Diego. Ph.D. from Penn State (2025) in hardware/software co-design for ML systems, with publications at ISCA, MICRO, HPCA, ICLR, NSDI, and PACT.https://cyanmishra92.github.io/en-ushttps://cyanmishra92.github.io/blog/hello-world/https://cyanmishra92.github.io/blog/hello-world/A quick note on what I'll write here, and what I won't.Sat, 25 Apr 2026 00:00:00 GMTblogmetahttps://cyanmishra92.github.io/news/#2026-01-15-joined-armhttps://cyanmishra92.github.io/news/#2026-01-15-joined-armStarted as a Performance and Power Engineer on the SoC Architecture Performance team at Arm in San Diego. Working on pre-silicon performance and power analysis for next-generation heterogeneous SoCs targeting Edge AI, IoT, and automotive woThu, 15 Jan 2026 00:00:00 GMTnewscareerarmhttps://cyanmishra92.github.io/news/#2025-08-15-phd-defensehttps://cyanmishra92.github.io/news/#2025-08-15-phd-defenseDefended my Ph.D. thesis in Computer Science and Engineering at Penn State and officially graduated. Seven years in the Microsystems Design Lab, advised by Mahmut Taylan Kandemir and Jack Sampson.Fri, 15 Aug 2025 00:00:00 GMTnewsphdmilestonepenn-statehttps://cyanmishra92.github.io/projects/salient-store/https://cyanmishra92.github.io/projects/salient-store/Salient Store — our work on FPGA-accelerated computational storage for continuous-learning edge servers — was accepted at PACT 2025.Tue, 01 Apr 2025 00:00:00 GMTnewspaperpactcomputational-storagehttps://cyanmishra92.github.io/news/#2025-02-15-ipdps-cordhttps://cyanmishra92.github.io/news/#2025-02-15-ipdps-cordCORD — parallelizing query processing across multiple computational storage devices — was accepted at IPDPS 2025. Joint work with Wahid Uz Zaman, Saleh AlSaleh, Abutalib Aghayev, and Mahmut Taylan Kandemir.Sat, 15 Feb 2025 00:00:00 GMTnewspaperipdpscomputational-storagehttps://cyanmishra92.github.io/projects/nexume/https://cyanmishra92.github.io/projects/nexume/NExUME — intermittency-aware DNN training that adapts dropout and quantization to harvested energy, with up to 22% accuracy gain at <5% compute overhead — was accepted at ICLR 2025.Wed, 22 Jan 2025 00:00:00 GMTnewspapericlrintermittent-computinghttps://arxiv.org/abs/2410.05435https://arxiv.org/abs/2410.05435Posted the Salient Store preprint to arXiv (2410.05435). FPGA-accelerated computational storage for continuous-learning edge servers — 6× throughput, 6.1× less data movement.Tue, 08 Oct 2024 00:00:00 GMTnewsarxivcomputational-storagehttps://arxiv.org/abs/2408.14379https://arxiv.org/abs/2408.14379Posted the Synergistic and Efficient Edge-Host Communication for Energy-Harvesting WSNs preprint to arXiv (2408.14379).Mon, 26 Aug 2024 00:00:00 GMTnewsarxiveh-wsncommunicationhttps://arxiv.org/abs/2408.13696https://arxiv.org/abs/2408.13696Posted the NExUME preprint to arXiv (2408.13696). Revisiting DNN training for intermittently-powered energy-harvesting micro-computers.Sun, 25 Aug 2024 00:00:00 GMTnewsarxivintermittent-computingnexumehttps://cyanmishra92.github.io/projects/usas/https://cyanmishra92.github.io/projects/usas/Presented Usás — a sustainable continuous-learning framework for solar-powered edge servers — at HPCA 2024 in Edinburgh.Mon, 04 Mar 2024 00:00:00 GMTnewspaperhpcasustainable-computinghttps://cyanmishra92.github.io/news/#2022-10-03-micro-point-cloudhttps://cyanmishra92.github.io/news/#2022-10-03-micro-point-cloudPresented our MICRO 2022 paper on pushing point cloud compression to the edge — a two-stage pipeline for geometry (parallel octree) and attribute compression with significant speedups and energy savings on edge platforms.Mon, 03 Oct 2022 00:00:00 GMTnewspapermicropoint-cloudhttps://youtu.be/VAsB1XBuRZ0https://youtu.be/VAsB1XBuRZ0Presented Cocktail at NSDI 2022 in Renton, WA — multidimensional optimization for model serving in the cloud. Recording on YouTube.Mon, 04 Apr 2022 00:00:00 GMTnewspapernsdicloudinference-servinghttps://cyanmishra92.github.io/projects/origin/https://cyanmishra92.github.io/projects/origin/Origin — on-device intelligence for human activity recognition using energy-harvesting wireless sensor networks — was nominated for Best Paper at DATE 2021. Joint work with Jack Sampson, Mahmut Taylan Kandemir, and Vijaykrishnan Narayanan.Mon, 08 Feb 2021 00:00:00 GMTnewsawarddateeh-wsnorigin