BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//juliacon-2026//speaker//HRBUXF
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-juliacon-2026-TVG9AR@pretalx.com
DTSTART;TZID=CET:20260814T120000
DTEND;TZID=CET:20260814T121500
DESCRIPTION:This talk presents JADE (Julia Analytics Decision Engine)\, a p
 roduction system serving a wide variety of companies\, that resolves BI de
 velopment tension through rapid model development alongside dynamic Julia 
 package generation\, intelligent precompilation\, and cache-first architec
 ture.\n\n  JADE generates complete Julia packages at runtime from domain-s
 pecific analytical models. Each generated package contains 10K+ lines of J
 ulia code implementing hundreds of analytical functions\, multi-dimensiona
 l data structures\, dependency graphs\, and model-specific formula chains.
  The system currently serves enterprise clients with production models pro
 cessing gigs of data and supporting thousands of function calls per evalua
 tion with minimal query latency for warm queries.\n\n  The core innovation
  is our three-stage precompilation strategy that balances compilation over
 head against runtime performance. First\, we precompile reusable function 
 templates once and distribute them via a shared depot\, covering hundreds 
 of analytical functions relevant to financial modeling\, statistical analy
 sis\, and multi-dimensional array operations.\n  Second\, when models chan
 ge\, we generate model-specific code and combine it with precompiled templ
 ates without triggering full recompilation. \n  Third\, we maintain a hash
 -based cache of compiled packages that delivers instant results for cache 
 hits while compiling updated packages in the background.\n\n  We support t
 wo deployment patterns with different performance characteristics. In loca
 l mode\, Julia processes run on workstations or servers\, achieving fast c
 old starts and near instant warm queries.\n  In Hub mode\, distributed Jul
 ia processes use shared Registries and caches\, delivering low latency que
 ries with horizontal scaling.\n\n  Performance optimization is central to 
 JADE's architecture. We employ automatic multi-threading for large array o
 perations with custom chunking strategies\, coordinate remapping systems t
 hat precompute dimension maps to avoid allocations in hot loops\, smart di
 rty state tracking via dependency graphs that reduces recomputation\, and 
 union splitting macros to limit reliance on type dispatch while maintainin
 g flexibility for heterogeneous data.\n\n  The talk will cover practical e
 ngineering challenges we solved: managing Julia depot paths across deploym
 ent environments\, implementing intelligent cache invalidation strategies\
 , optimizing precompilation workloads with custom compile statements\, han
 dling package versioning and upgrades in production\, along with debugging
  performance issues in generated code. We'll share performance measurement
 s\, code examples\, and lessons learned from a year of production deployme
 nt.\n\n  This work demonstrates Julia's readiness for enterprise-critical 
 systems and provides an inspiration for organizations building dynamic cod
 e generation platforms and formula chain engines. The techniques we presen
 t—precompilation strategies\, caching architectures\, and performance op
 timization patterns—are broadly applicable to any domain requiring flexi
 ble\, high-performance analytics.\n  Our experience shows that Julia's com
 bination of performance\, metaprogramming capabilities\, and ecosystem mat
 urity enables production systems that were previously impractical.\n\n  Ta
 rget audiences include enterprise developers integrating Julia into busine
 ss intelligence platforms\, developers building code generation systems fo
 r domain-specific languages\, performance-focused Julia users\, and organi
 zations evaluating Julia for production analytics workloads.
DTSTAMP:20260710T083415Z
LOCATION:Alte Mensa — Audi Max
SUMMARY:BI Engine in Julia - Matthew Muyres / Chase Cowart
URL:https://pretalx.com/juliacon-2026/talk/TVG9AR/
END:VEVENT
END:VCALENDAR