Subject: Seminars relevant to Informatics
List archive
ICSA Visitor Seminar - Wednesday 28th October 11:00, Room G03, Speaker: Dr Anton Lokhmotov (University of Cambridge)
- From: Andrew Macnaughtan <andrew.macnaughtan AT ed.ac.uk>
- To: "inf-seminars AT mlist.is.ed.ac.uk " <inf-seminars AT mlist.is.ed.ac.uk>, members ICSA <icsa-members AT inf.ed.ac.uk>
- Subject: ICSA Visitor Seminar - Wednesday 28th October 11:00, Room G03, Speaker: Dr Anton Lokhmotov (University of Cambridge)
- Date: Thu, 22 Oct 2026 15:36:44 +0000
BEGIN:VCALENDAR METHOD:REQUEST PRODID:Microsoft Exchange Server 2010 VERSION:2.0 BEGIN:VTIMEZONE TZID:GMT Standard Time BEGIN:STANDARD DTSTART:16010101T020000 TZOFFSETFROM:+0100 TZOFFSETTO:+0000 RRULE:FREQ=YEARLY;INTERVAL=1;BYDAY=-1SU;BYMONTH=10 END:STANDARD BEGIN:DAYLIGHT DTSTART:16010101T010000 TZOFFSETFROM:+0000 TZOFFSETTO:+0100 RRULE:FREQ=YEARLY;INTERVAL=1;BYDAY=-1SU;BYMONTH=3 END:DAYLIGHT END:VTIMEZONE BEGIN:VEVENT ORGANIZER;CN=Andrew Macnaughtan:mailto:andrew.macnaughtan AT ed.ac.uk ATTENDEE;ROLE=REQ-PARTICIPANT;PARTSTAT=NEEDS-ACTION;RSVP=TRUE;CN=inf-semina rs AT mlist.is.ed.ac.uk :mailto:inf-seminars AT mlist.is.ed.ac.uk ATTENDEE;ROLE=REQ-PARTICIPANT;PARTSTAT=NEEDS-ACTION;RSVP=TRUE;CN=members IC SA:mailto:icsa-members AT inf.ed.ac.uk DESCRIPTION;LANGUAGE=en-GB:Good afternoon\,\n\nAn ICSA visitor seminar will be taking place on Wednesday 28th October at 11am in room G.03\nDr Anton Lokhmotov (University of Cambridge) will be speaking.\n\nIf anyone wishes to meet Dr Lokhmotov please contact Boris Gort - Boris.Grot AT ed.ac.uk\n\nTi tle: Optimizing AI Infrastructure: Systematically\, Agentically\, Collecti vely\nMaking AI software fit AI hardware like a glove\n\nAbstract: AI infr astructure is the hardware and software that AI runs on: specialized serve rs with accelerator chips\, memories\, networks\, and frameworks\, runtime s\, compilers\, kernels that turn models into computation. AI Infrastructu re is essential to both training and deploying AI\, and it largely determi nes how fast\, how expensive\, and how energy-hungry the result is.\nMakin g AI software fit AI hardware like a glove is hard. There are never enough experts or enough time to achieve the most ambitious and effective optimi zation across diverse workloads\, target hardware\, and user-specific perf ormance\, quality\, and cost requirements.\nAt KRAI\, we are working with leading companies to automate low-level optimization through systematic ag entic engineering. In one striking case\, we accelerated a reference imple mentation of a frontier open-source model by over 1\,000 times in under a week. In this talk\, I will share several examples\, and what they taught us: the good\, the bad\, and the ugly of putting AI agents to work on real optimization problems.\nThese experiences underpin The Collective\, our f ramework of cooperating specialized agents that tirelessly work together t o optimize AI infrastructure automatically. I will describe how it is desi gned\, where agents already help\, where they still fall short\, and which open problems are worth tackling. Attendees will leave with a practical s ense of what agentic optimization can and cannot do today\, and where they could contribute.\n\nBrief Bio: Dr Anton Lokhmotov has spent over 20 year s designing and optimizing computer systems as an entrepreneur\, engineer\ , and researcher. In 2020\, he founded KRAI to build ultra-efficient\, cos t-effective systems for Artificial Intelligence applications. Previously\, he co-founded dividiti (dv/dt) and led GPU Compute (OpenCL) compiler deve lopment at Arm.\nKRAI leverages the team’s decades of expertise in autom ated optimization to reduce the total cost of ownership (TCO) and mitigate risks of deploying AI solutions in enterprise.\n\nAs a Founding Member of MLCommons' MLPerf benchmarking competitions\, the KRAI team has acted as an "Olympic coach" to industry leaders such as Qualcomm\, Google\, Nebius\ , HPE and Dell\, contributing to some of the fastest and most energy-effic ient results in MLPerf history.\nDr Lokhmotov holds a PhD in Computer Scie nce from the University of Cambridge\, where he is an External Member of t he Faculty Board.\n\nKind regards\,\n\nAndrew\n\nThe University of Edinbur gh is a charitable body\, registered in Scotland\, with registration numbe r SC005336. Is e buidheann carthannais a th’ ann an Oilthigh Dhùn Èide ann\, clàraichte an Alba\, àireamh clàraidh SC005336.\n UID:040000008200E00074C5B7101A82E00800000000D7E218243B62DD01000000000000000 0100000008F381438AD0FD24AA342196C02B5D021 SUMMARY;LANGUAGE=en-GB:ICSA Visitor Seminar - Wednesday 28th October 11:00\ , Room G03\, Speaker: Dr Anton Lokhmotov (University of Cambridge) DTSTART;TZID=GMT Standard Time:20261028T110000 DTEND;TZID=GMT Standard Time:20261028T120000 CLASS:PUBLIC PRIORITY:5 DTSTAMP:20261022T153644Z TRANSP:OPAQUE STATUS:CONFIRMED SEQUENCE:1 LOCATION;LANGUAGE=en-GB:G.03\, Informatics Forum X-MICROSOFT-CDO-APPT-SEQUENCE:1 X-MICROSOFT-CDO-OWNERAPPTID:2125116858 X-MICROSOFT-CDO-BUSYSTATUS:TENTATIVE X-MICROSOFT-CDO-INTENDEDSTATUS:BUSY X-MICROSOFT-CDO-ALLDAYEVENT:FALSE X-MICROSOFT-CDO-IMPORTANCE:1 X-MICROSOFT-CDO-INSTTYPE:0 X-MICROSOFT-DONOTFORWARDMEETING:FALSE X-MICROSOFT-DISALLOW-COUNTER:FALSE X-MICROSOFT-REQUESTEDATTENDANCEMODE:DEFAULT X-MICROSOFT-ISRESPONSEREQUESTED:TRUE X-MICROSOFT-LOCATIONDISPLAYNAME:G.03\, Informatics Forum X-MICROSOFT-LOCATIONSOURCE:None X-MICROSOFT-LOCATIONS:[{"DisplayName":"G.03\, Informatics Forum"\,"Location Annotation":""\,"LocationUri":""\,"LocationStreet":""\,"LocationCity":""\, "LocationState":""\,"LocationCountry":""\,"LocationPostalCode":""\,"Locati onFullAddress":""}] BEGIN:VALARM DESCRIPTION:REMINDER TRIGGER;RELATED=START:-PT15M ACTION:DISPLAY END:VALARM END:VEVENT END:VCALENDAR
-
ICSA Visitor Seminar - Wednesday 28th October 11:00, Room G03, Speaker: Dr Anton Lokhmotov (University of Cambridge),
Andrew Macnaughtan, 10/22/2026
- <Possible follow-up(s)>
- ICSA Visitor Seminar - Wednesday 28th October 11:00, Room G03, Speaker: Dr Anton Lokhmotov (University of Cambridge), Informatics Institute News, 10/22/2026
Archive powered by MHonArc 2.6.24.