Research Group Leader at MPI-SWS

Laurent
Bindschaedler

I lead the Data Systems Group (DSG) at the Max Planck Institute for Software Systems.

I build runtime guarantees for AI agents that act on real systems: when an action becomes permanent, what state an agent must keep, and how to show afterward what happened.

This work builds on my earlier research in operating systems, databases, and distributed systems at EPFL and MIT: skew-tolerant analytics, disaggregated storage, real-time graph mining, and learned systems.

Prev: Postdoctoral Fellow at MIT CSAIL · Ph.D. at EPFL

Prospective Students

CV Research Statement

Research Interests

Agentic AI Infrastructure

Data Systems for AI

Trustworthy AI

Learned Systems

Distributed Systems

Operating Systems

Latest Blog Posts

NewsSep 2, 2026

Two Papers at the ASE 2026 Workshops in Munich

ForkOps at SWGeno 2026 and Whose Oracle Is It? at APR 2026, both workshops co-located with ASE in Munich. Generated software needs fleet operations, and generated patches need an authority problem solved.

NewsAug 31, 2026

Three Papers at the SOSP 2026 Workshops, and a Keynote at AgenticOS

Two papers at AgenticOS 2026 and one at PACMI 2026, both workshops co-located with SOSP in Prague. And I will be giving the keynote at AgenticOS. If you are attending either workshop, come say hi.

NewsJul 31, 2026

Two Papers Accepted at BCCA 2026

Two new papers with Quentin Botha and Christoph Siebenbrunner at BCCA 2026 in Barcelona. Agent Flight Recorder gives long-horizon agents a tamper-evident audit trail anyone can verify, and Bonded Recourse settles the harm an agent leaves behind through smart contracts.

OpinionJul 4, 2026

The Conservation of Complexity: The Ends Still Need a Middle

Everyone is predicting that AI will flatten the software stack into one English-speaking agent, until nothing of the old software remains. It will not go that far. Complexity is conserved, the end-to-end principle still collects its debt, and the stack does not flatten all the way down. It sorts itself by what you can afford to trust.

NewsJun 12, 2026

The Invisible Lottery: Algorithm Steering in LLM Code Generation at ICML 2026

New paper with Akanksha Narula and Mofasshara Rafique at ICML 2026 in Seoul. Incidental prompt cues silently decide which algorithm an LLM ships, even when every output passes the same tests.

NewsApr 20, 2026

What Survives When Code Doesn't? Keynote at EuroMLSys 2026

AgenticOS Workshop at NeurIPS 2026(Non-archival)

Agent Flight Recorder: Tamper-Evident Audit Trails with On-Chain Anchoring for Long-Horizon Tool-Using Agents

2026

Laurent Bindschaedler, Quentin Botha, Christoph Siebenbrunner

BCCA 2026

PDF

View All Publications

Research Projects

Agentic AI Infrastructure

Agent Runtime Contracts

AI agents now edit code, update databases, and restart services, and many of their failures happen in the runtime around the model. We define and enforce four contracts between the model and the systems it acts on. Effect: an external effect becomes permanent only when the runtime commits it (Atomix, Externalization Barriers, Irreversibility Budget, typed recovery actions). State: agent memory is managed like virtual memory (ClawVM, ClawMMU). Composition: an approval stays valid only while the evidence it relied on is unchanged. Intent and evidence: the runtime records each model call in tamper-evident form, so operators can check afterward which context shaped an action (Agent Flight Recorder).

Trustworthy AI

Trustworthy Code Generation

Incidental cues in a prompt can change which algorithm a model chooses by up to 100 percentage points, among solutions that pass the same tests. Standard benchmarks, which check only outcomes, cannot detect this. We study how context steers generated code, who decides what counts as correct when agents repair programs, and how to build LLM pipelines for code and SQL that can be checked.

Data Systems for AI

Cache Saver

Cache Saver caches LLM calls in multi-step pipelines. Its namespace-aware, list-valued cache keeps sampled outputs statistically valid and makes experiments reproducible. It cuts inference cost by about 25% and CO2 emissions by about 35% on average.

Data Systems for AI

LLM-Enhanced Databases

IOLM-DB compiles recurring semantic operators over table columns into specialized models that are 1.8 to 2.8 times smaller in memory than the general-purpose model. Our semantic cache for OLAP canonicalizes SQL and natural-language queries with an LLM and reaches an 82% hit rate with zero false hits.

Graph Systems

Large-Scale GNN Training

Grappa trains graph neural networks by exchanging only gradients between partitions and periodically repartitions the graph to expose new neighborhoods. In the paper's experiments it trains 4x faster on average (up to 13x) than the baseline systems, and it scales to trillion-edge graphs on commodity hardware.

Experience

Nov 2022 – present

Research Group Leader

Max Planck Institute for Software Systems (MPI-SWS)

Sep 2020 – Nov 2022

Postdoctoral Fellow

MIT CSAIL

Sep 2015 – Sep 2020

Doctoral Assistant

EPFL, LABOS

Mar 2014 – Aug 2015

Research Assistant

EPFL, LABOS

Jul 2013 – Feb 2014

IT Consultant – Acting Head of IT

Lobnek Wealth Management

Aug 2011 – Jun 2013

Co-founder & VP Engineering

LakeMind

Jan 2011 – Jul 2011

Intern

Nokia Research Center

Jul 2006 – Dec 2018

IT Pioneer, Chief of Staff, 1st Army Region

Swiss Armed Forces

Education

Ph.D. in Computer Science

EPFL, Lausanne, Switzerland

2015 – 2020

Thesis: An Architecture for Load Balance in Computer Cluster Applications
Advisor: Prof. Willy Zwaenepoel

M.S. in Computer Science

EPFL, Lausanne, Switzerland

2009 – 2011

Thesis: Track Me If You Can
Advisor: Prof. Jean-Pierre Hubaux

B.S. in Computer Science

EPFL, Lausanne, Switzerland

2006 – 2009

Awards & Honors

Swiss AI Initiative Grant (Co-Applicant)

CLINIVERSE auditable clinical knowledge base, CSCS Alps (50,000 GPU-hours)

2026

Gold Reviewer

ICML 2026

2026

Associate Fellow

Saarland University

2025

Two Research Grants (EUR 80,000 each)

armasuisse (Swiss defense procurement), SkyPulse satellite data system

2023, 2024

Early Postdoc.Mobility Fellowship

Swiss National Science Foundation

2020

Best Presentation Award

ASPLOS 2020

2020

Teaching Assistant Award

EPFL IC Faculty

2017

Public Prize Winner

Exposure Science Movie Hackathon

2017

Ph.D. Fellowship

EPFL EDIC

2015

Venture Kick 1st Round

LakeMind

2012

LaurentBindschaedler

Research Interests

Agentic AI Infrastructure

Data Systems for AI

Trustworthy AI

Learned Systems

Distributed Systems

Operating Systems

Latest Blog Posts

Two Papers at the ASE 2026 Workshops in Munich

Three Papers at the SOSP 2026 Workshops, and a Keynote at AgenticOS

Two Papers Accepted at BCCA 2026

The Conservation of Complexity: The Ends Still Need a Middle

The Invisible Lottery: Algorithm Steering in LLM Code Generation at ICML 2026

What Survives When Code Doesn't? Keynote at EuroMLSys 2026

Latest Publications

Ghost Tool Calls: Issue-Time Privacy for Speculative Agent Tools

When to Cross the Rubicon: Settlement Reliability for LLM Agent Tool Use

The Working Set of a Coding Agent: Coherence Debt in Repository-Scale Tasks

ClawMMU: Protection-Coherent Virtual Memory for Stateful LLM Agents

Agent Flight Recorder: Tamper-Evident Audit Trails with On-Chain Anchoring for Long-Horizon Tool-Using Agents

Research Projects

Agent Runtime Contracts

Trustworthy Code Generation

Cache Saver

LLM-Enhanced Databases

Large-Scale GNN Training

Benchmarking Learned Systems

Teaching

Operating Systems (OS2026)

Operating Systems (OS2024)

Systems for Large Language Models (SysLLM)

Distributed Systems (DS2023)

Experience

Research Group Leader

Postdoctoral Fellow

Doctoral Assistant

Research Assistant

IT Consultant – Acting Head of IT

Co-founder & VP Engineering

Intern

IT Pioneer, Chief of Staff, 1st Army Region

Education

Ph.D. in Computer Science

M.S. in Computer Science

B.S. in Computer Science

Awards & Honors

Swiss AI Initiative Grant (Co-Applicant)

Gold Reviewer

Associate Fellow

Two Research Grants (EUR 80,000 each)

Early Postdoc.Mobility Fellowship

Best Presentation Award

Teaching Assistant Award

Public Prize Winner

Ph.D. Fellowship

Venture Kick 1st Round

Laurent
Bindschaedler