Northbound AI Summit 2026

31 speakers

Sofía ÁlvarezStaff Product Engineer · RelayWorksSofía ships agent products at RelayWorks and owns the production feedback loop after launch. She documents incidents where technically successful agents optimized the wrong outcome, then turns each failure into a durable guardrail.1 sessionZAZara AminDocument Intelligence Lead · Papertrail HealthZara leads document intelligence at Papertrail Health, processing forms and clinical records where extraction must be traceable to source pixels. Her team combines layout models, vision systems, and human-verifiable grounding.1 sessionJBJordan BellCompute Economist · Capacity IndexJordan analyzes accelerator supply, inference pricing, and cloud capacity contracts at Capacity Index. Their research translates opaque provider markets into planning models used by engineering and finance leaders.1 sessionLBLuca BianchiML Systems Engineer · Piccolo AILuca builds compact language models and the training systems behind them at Piccolo AI. He has helped teams move narrow production tasks from frontier APIs to fine-tuned small models while preserving measurable quality.1 sessionNBNia BrooksApplied AI Lead · SchemaWorksNia leads applied AI at SchemaWorks, where structured generation replaced dozens of fragile extraction pipelines. Her work covers constrained decoding, semantic validation, repair strategies, and schema design that models can reliably satisfy.1 sessionMaya ChenVP, AI Platform · Northstar SystemsMaya leads the AI platform organization at Northstar Systems, where her teams operate shared inference, retrieval, and evaluation services for hundreds of product engineers. She previously built distributed systems at two developer infrastructure companies.1 sessionSamira ColeStaff Engineer · LatticeworkSamira leads the retrieval and memory platform at Latticework, where her team operates one of the larger production assistant fleets outside the model labs. Before that she spent six years building developer-tools infrastructure. She speaks about the unglamorous parts of AI engineering — queues, evals, and the bill.2 sessionsIDInês DuarteDeveloper Infrastructure Fellow · Local Cloud FoundationInês is a fellow at the Local Cloud Foundation studying infrastructure that assembles itself around application code. She previously led developer experience teams and now advises open-source projects on platform ergonomics.1 sessionADArthur DuboisEditor · Practical AI ReviewArthur edits Practical AI Review, where he turns production case studies into deeply reported technical features. He has moderated engineering panels on five continents and is known for insisting on numbers behind every prediction.1 sessionOEOmar El-AminOpen Models Lead · Meridian ComputeOmar leads open-model deployment at Meridian Compute and has taken language models from research checkpoints to regulated production workloads. His work focuses on serving economics, reproducible evaluation, and the operational trade-offs behind model ownership.1 sessionDana FieldsDana chairs the program committee at DevFlow Conf and has read north of four thousand CFP submissions across eight years of programs. She moderates panels the way she runs review meetings — with a timer and zero patience for product pitches — and writes about program design and first-time-speaker development.1 sessionNoor HaddadNoor is VP of Engineering at the Post-SaaS Institute, where she studies how teams replace subscription software with open, self-hosted alternatives — and what it costs them. She previously ran platform engineering at two infrastructure companies and writes the Post-SaaS Notes newsletter.1 sessionLHLayla HassanVP of Engineering · Harbor FinanceLayla is VP of Engineering at Harbor Finance, where she moved a regulated AI platform from a vendor bundle to an internal stack and later moved selected layers back. She speaks candidly about ownership cost and organizational fit.1 sessionIKIsaac KimCapacity Strategy Lead · Tessellate CloudIsaac leads accelerator capacity strategy at Tessellate Cloud and negotiates reserved, on-demand, and spot portfolios for large inference fleets. He turns utilization traces and product forecasts into procurement decisions.1 sessionBLBenjamin LiuPrincipal Engineer · Cacheline AIBenjamin is a principal engineer at Cacheline AI focused on safe caching for model-backed products. He has operated exact, prefix, retrieval, and semantic caches and keeps a detailed archive of the invalidation incidents each one caused.1 sessionRohan MehtaFounder · Stagecraft Open SourceRohan founded Stagecraft Open Source to give community conferences fast, maintainable program sites without platform fees. He works across data modeling, web performance, accessibility, and the operational reality of event-day edits.1 sessionAMAdwoa MensahModel Operations Director · Sankofa ComputeAdwoa runs model operations at Sankofa Compute, where her team serves open-weight models for products across Africa and Europe. She focuses on reliability, multilingual quality, and the staffing cost that rarely appears in benchmark comparisons.1 sessionAlex MoreauDeveloper Advocate · HarborlineAlex is a Developer Advocate at Harborline, where they maintain the open-source retrieval toolkit ragkit and write a long-running field-notes series on search quality. They co-host the Retrieval Roundtable podcast and have taught retrieval workshops at a dozen developer conferences.1 sessionPriya NarayananPrincipal Security Researcher · Boundary LabsPriya researches how language-model applications fail at trust boundaries, with a focus on indirect prompt injection and tool authorization. At Boundary Labs she partners with product teams to turn red-team findings into practical, layered controls.1 sessionTNTomáš NovákObservability Architect · TracegardenTomáš builds tracing and online-evaluation systems for language-model applications at Tracegarden. He specializes in making latency, token cost, tool calls, and semantic quality visible in one production trace.1 sessionAmina OkaforDirector of ML Efficiency · Common ThreadAmina owns model efficiency at Common Thread, where she reduced inference cost while expanding the company model portfolio. Her team measures every optimization against held-out quality and publishes an internal tokens-per-dollar index.1 sessionMOMichelle OkoroExecutive Director · Speaker FirstMichelle runs Speaker First and has studied completion and acceptance patterns across years of conference calls for proposals. She helps program teams shorten forms, support first-time speakers, and make review decisions more transparent.1 sessionLena OrtizLena leads deliverability at Inbox Works, helping event teams keep transactional and announcement email out of the spam folder. She has run SPF, DKIM, and DMARC rollouts across fleets of sending domains, and her hands-on authentication workshop is built from those migrations.1 sessionEPElena PetrovaResearch Engineering Director · Gradient CommonsElena directs research engineering at Gradient Commons, where her group evaluates state-space, recurrent, sparse, and hybrid model architectures under matched training budgets. She focuses on evidence that production teams can act on.1 sessionEREli RosenbergEvaluation Engineer · Proofpoint AIEli designs evaluation systems at Proofpoint AI, from golden datasets and deterministic graders to calibrated language-model judges. They help product teams translate support incidents into regression cases that can block a release.1 sessionKSKenji SatoOn-device ML Engineer · Pocket ModelsKenji builds local-first intelligence at Pocket Models, spanning on-device inference, private retrieval, and graceful server fallback. He tests every demonstration offline so the architecture cannot hide a network dependency.1 sessionMarco SilvaMarco is a Platform Engineer at SwarmScale, where he runs agent swarms that peak at several thousand concurrent runs on a D1-backed control plane. He spends most days reading query plans and traces, and shares the findings on the SwarmScale engineering blog.2 sessionsYuki TanakaSearch Infrastructure Lead · Kintsugi DataYuki leads search infrastructure at Kintsugi Data, combining lexical retrieval, embeddings, structured filters, and reranking in high-volume products. She writes about the evaluation failures that force retrieval teams to rebuild.1 sessionMTMalik ThompsonDatabase Reliability Engineer · Edge LedgerMalik is a database reliability engineer at Edge Ledger and has guided dozens of live D1 schema changes. He teaches expand-and-contract migrations, bounded backfills, and verification queries that make edge-database cutovers boring.1 sessionGWGrace WuProgram Director · Systems ForumGrace directs the Systems Forum conference program and has built technical agendas across infrastructure, databases, and applied AI. She is known for closing panels that turn three days of claims into specific, testable predictions.1 session