In the tenth and final tutorial of the Mastering MLflow for GenAI series, Jules Damji (Databricks) builds a multi-agent supervisor pattern with LangGraph and MLflow, routing queries to specialized agents, synthesizing results, and evaluating the full workflow with granular traces and scorers.
This session simulates a Databricks Agent framework–style supervisor using LangGraph’s StateGraph for acyclic, state-machine-like execution. The use case is a FEMA disaster-response assistant over roughly simulated records.
What You’ll Learn:
🔹 Multi-agent supervisor pattern: a supervisor fans out queries to specialized agents, then returns one cohesive answer
🔹 Supervisor router (Node 1): classify natural-language queries and route to a Genie-like agent, a knowledge assistant, or both
🔹 Genie-like agent (structured data): Natural Language → SQL with an LLM chain, executed against FEMA data via Python Pandas + SQLite
🔹 Knowledge-based assistant (unstructured data): document retrievers, OpenAI embeddings, and cosine similarity over in-memory files like PDFs
🔹 Synthesizer (Node 3): combine one or both agent outputs into a single response
🔹 Granular MLflow tracing: define agent tools: query, tool, embedding, and retriever functions with MLflow span markers
🔹 Built-in scorers: use MLflow’s 60+ scorers for answer relevance, safety, and guideline adherence on an evaluation dataset
🔹 Custom routing accuracy: implement a customer MLflow @scorer metric that compares supervisor fan-out against ground truth
🔹 Extra practice: supplementary notebooks for Deep Agents and CrewAI multi-agent frameworks
Resources:
🔗 Notebook 1.10: https://github.com/dmatrix/mlflow-genai-tutorials/blob/main/10_multi_agent_supervisor.ipynb
🎥 Full Series Playlist: https://youtube.com/playlist?list=PLaoPu6xpLk9EI99TuOjSgy-UuDWowJ_mR
The Hugging Face research team discusses the paper "Understanding Reasoning from Pretraining to Post-Training" which discovers a joint scaling law for pre-training and reinforcement learning.
Paper link: https://huggingface.co/papers/2607.16097
Dov Katz (Managing Director at Morgan Stanley) presents a deep-dive walkthrough of the Risk Navigator tool built for the OSERA (Open Source Enterprise Resiliency Alliance) initiative. He demonstrates how platform engineering and security teams can parse complex Software Bills of Materials (SBOMs), prioritize back-patching requirements across thousands of repositories, and use automated OpenRewrite pipelines to burn down transitive CVE vulnerabilities at enterprise scale.
🗽 Catch Us in New York! Ready to secure your software supply chain and scale automated patch management? Join the enterprise security community at OSFF New York on November 4–5, 2026.
🎟️ Register Now: https://hubs.ly/Q04n_bZL0
🔥 20% OFF DISCOUNT CODE: 26YTOSFFNY20C
🕒 Timestamps:
0:00 Unconference Kickoff: Why Banks Are Commissioning Shared Back-Patches
1:00 Navigating Shared Infrastructure Priorities and Commissioned Capacity
1:39 Introducing Risk Navigator: Mapping Enterprise Dependency Data Sets
2:30 Quantifying Productivity: Using Reduced CVE/CWE Counts over Merged PRs
3:35 Navigating Transitive Dependencies: High-Leverage Upgrades vs. Developer Friction
4:20 Spec-Driven Local Deployments: Tailoring Risk Navigator for Corporate Estates
5:30 Dashboard Walkthrough: Filtering Patch-Level Opportunities and Version Offsets
6:10 Setting CVSS Vulnerability Thresholds and Evaluating Active Exploits
7:20 Identifying "Dead Ends": When Software Requires Commissioned Back-Patching
8:00 The Version-Bumping Shopping Cart: Generating OpenRewrite YAML Automations
8:41 Back-Patch Priority Matrices: Mutualizing Morgan Stanley, Citi, and Deutsche Bank Lists 9:30 Targeting Amplifiers: Base Container Images and Transitive Spring Framework CVEs
10:00 Privacy and Execution: Running Static JSON Data Pipelines Privately Behind Firewalls
10:40 Transitioning to Scale: Remediation Workflows with Moderne
📊 The Problem: The Transitive Dependency Patching Nightmare
When security scanners flag tens of thousands of Common Vulnerabilities and Exposures (CVEs) across an enterprise codebase containing 60,000 source repositories, platform teams face paralysis. Sending thousands of individual developers to manually add transitive overrides creates massive technical debt that becomes impossible to unwind when official patches release. Furthermore, many critical enterprise dependencies hit "dead ends"—where upstream maintainers have stopped supporting legacy versions—leaving banks with high-risk vulnerabilities and no clear upgrade path.
🏗️ The Solution: Risk Navigator & OpenRewrite Mutualization
Dov Katz demonstrates how financial institutions can use Spec-Driven decision support tools to prioritize remediation:
High-Leverage Amplifiers: Locating core framework and base container image dependencies (such as Spring or base Linux RPMs) where a single patch closes thousands of downstream vulnerability touchpoints simultaneously.
Dead-End Identification: Distinguishing between software that can be immediately bumped to minor/patch versions versus legacy packages requiring joint industry-commissioned back-patching.
Automated Shopping Carts: Selecting vetted dependency targets and exporting OpenRewrite YAML scripts to automatically execute version updates across thousands of repositories.
⚙️ Why This Matters for Financial Engineering
Quantifying True Security ROI: Shifting engineering metrics away from vanity signals (like merged PR counts) to measurable security burn-down rates (reducing active CVE touchpoints from 500,000 to 100,000).
Local Data Privacy: Running static, spec-driven analysis engines completely offline inside corporate environments without sending internal dependency trees or proprietary code to external APIs.
🌐 More about FINOS: https://www.finos.org/
📧 Join our newsletter: https://www.finos.org/sign-up
🎙️ Listen to our Open Source in Finance Podcast: https://www.youtube.com/@FINOS/podcasts
LinkedIn: https://www.linkedin.com/company/finosfoundation
Support us on Patreon and get an ad-free RSS feed with some early episodes. https://www.patreon.com/LateNightLinux
Initial Steam Machine reviews are broadly positive, Jellyfin is losing some key project leaders, Sony deletes more movies that people "bought", why you probably shouldn’t wipe your phone when the US authorities ask for access, Graham repurposes some discarded crypto hardware to play arcade games, Félim gets that old game working, a great terminal music player, and more.
Salling Group, Northern Europe’s largest retailer, shares how a modern SAP landscape is helping support more than 2,100 stores across six countries. Learn how SAP S/4HANA Cloud, RISE with SAP, and a strong supply chain foundation are helping the company improve efficiency, support employees, and better serve customers.
Read the article: https://news.sap.com/?p=246507
00:00 – The Autonomous Enterprise
00:23 – Who is Salling Group?
00:40 – Why AI Transformation?
00:58 – SAP S/4HANA Cloud and RISE with SAP
01:31 – What’s next for Salling Group?
Iranian forces have once again targeted Amazon AWS data centers in the Middle East, causing significant damage. Satellite images reveal burn scars and fires at multiple facilities. These attacks highlight the vulnerability of cloud infrastructure in the region. #CyberSecurity #AmazonAWS #MiddleEastConflict #DataCenter
You can find more about TWiT and subscribe to our full shows at https://podcasts.twit.tv/
Subscribe: https://twit.tv/subscribe
Products we recommend: https://www.amazon.com/shop/twitnetcastnetwork
TWiT may earn commissions on certain products.
Join our TWiT Community on Discourse: https://www.twit.community/
One developer has already built more than 100 web apps for Meta Ray-Ban Display. 👓
We’re talking a sky-gazing overlay that tracks constellations, EV controls, smart home lights you can adjust with a pinch gesture, and a racing game you steer by tilting your head.
In ‘Through the Lens’ episode 2, Herald breaks down how he built all of it and shares the technical playbook so you can start building your own. 🚀
🔗 Full episode and links to get started in the comments.
Video by CNCF [Cloud Native Computing Foundation] via YouTube
Running virtual machines on Kubernetes with KubeVirt works well until you need to live-migrate one. When a VM moves to another node, its networking is what breaks first: the IP address can change, established TCP sessions drop, and routes have to reconverge before traffic finds the VM again. This session shows how Calico handles all three so a VM can move between nodes while keeping its address and staying online.
We’ll start with KubeVirt’s networking modes and make the case for bridge mode, where the virt-launcher pod IP is the VM IP. That’s the foundation everything else builds on. From there, we cover the three parts of the problem. IP persistence is where Calico’s IPAM holds the VM’s address steady across the move. TCP session continuity is where an open connection survives the handover without reconnecting. Route convergence is where Felix and BGP pick up the gratuitous ARP and reprogram routes to the new node.
Each part is paired with a live demo on a multi-node Calico cluster, so you can watch the IP stay put, the TCP session hold, and cutover downtime stay near zero.
By the end, you’ll understand why live migration breaks VM networking on Kubernetes, and how Calico keeps the address while dropping the downtime.
I’ve been working on an app to scratch my itch for working out and tracking it, but needed something simple that I could also use as a reference each time I start a new workout.
=== Links ===
Show Notes:
https://wiki.opensourceisawesome.com/books/get-healthy/page/install-get-healthy
Get the AwesomeOpenSource Merchandise
https://osia.printify.me
Support my Channel and ongoing efforts through Patreon:
https://patreon.com/awesomeopensource
Buy Me a Coffee or Beer
https://paypal.me/BrianMcGonagill?country.x=US&locale.x=en_US
=== Get My Novel ===
Entanglement – by Brian McGonagill
Dave is just a normal guy with a normal job that he doesn’t like that much. That is, until he meets Leo, and gets the job he never imagined. Action, adventure, betrayal, love, loss, and some serious weirdness happen throughout this science fiction thrill-ride.
— eBook on Amazon by Country —
US – United States – https://www.amazon.com/dp/B0DZJ5PMHG
UK – United Kingdom – https://www.amazon.co.uk/dp/B0DZJ5PMHG
DE – Germany – https://www.amazon.de/dp/B0DZJ5PMHG
FR – France – https://www.amazon.fr/dp/B0DZJ5PMHG
ES – Spain – https://www.amazon.es/dp/B0DZJ5PMHG
IT – Italy – https://www.amazon.it/dp/B0DZJ5PMHG
NL – Netherlands – https://www.amazon.nl/dp/B0DZJ5PMHG
JP – Japan – https://www.amazon.co.jp/dp/B0DZJ5PMHG
BR – Brazil – https://www.amazon.com.br/dp/B0DZJ5PMHG
CA – Canada – https://www.amazon.ca/dp/B0DZJ5PMHG
MX – Mexico – https://www.amazon.com.mx/dp/B0DZJ5PMHG
AU – Australia – https://www.amazon.com.au/dp/B0DZJ5PMHG
IN – India – https://www.amazon.in/dp/B0DZJ5PMHG
— Paperback by Country —
US – United States – https://www.amazon.com/dp/B0DZPM682N
UK – United Kingdom – https://www.amazon.co.uk/dp/B0DZPM682N
DE – Germany – https://www.amazon.de/dp/B0DZPM682N
FR – France – https://www.amazon.fr/dp/B0DZPM682N
ES – Spain – https://www.amazon.es/dp/B0DZPM682N
IT – Italy – https://www.amazon.it/dp/B0DZPM682N
NL – Netherlands – https://www.amazon.nl/dp/B0DZPM682N
PL – Poland – https://www.amazon.pl/dp/B0DZPM682N
SE – Sweden – https://www.amazon.se/dp/B0DZPM682N
JP – Japan – https://www.amazon.co.jp/dp/B0DZPM682N
CA – Canada – https://www.amazon.ca/dp/B0DZPM682N
AU – Australia – https://www.amazon.com.au/dp/B0DZPM682N
— Hard cover by Country —
US – United States – https://www.amazon.com/dp/B0DZPVSRNZ
UK – United Kingdom – https://www.amazon.co.uk/dp/B0DZPVSRNZ
DE – Germany – https://www.amazon.de/dp/B0DZPVSRNZ
FR – France – https://www.amazon.fr/dp/B0DZPVSRNZ
ES – Spain – https://www.amazon.es/dp/B0DZPVSRNZ
IT – Italy – https://www.amazon.it/dp/B0DZPVSRNZ
NL – Netherlands – https://www.amazon.nl/dp/B0DZPVSRNZ
PL – Poland – https://www.amazon.pl/dp/B0DZPVSRNZ
SE – Sweden – https://www.amazon.se/dp/B0DZPVSRNZ
=== Timestamps ===
00:00 Beginning
00:09 Intro to Get Healthy
00:55 Trying out AI to Build – The Process
01:41 Support the Channel – Thank you Patrons
02:25 Checking th AI Code and Iterating
03:50 The App Overview
04:10 Tracking Measurements
04:45 System Sttings
05:10 User Profile
05:30 Back to Measurements and Adding Body Photos
07:35 Templates and Routines
08:40 Starting a Workout
12:50 Mobile View
14:45 Scheduling and Reports
15:25 Exporting your Data or Deleting Data Completely
We use cookies on our website to give you the most relevant experience by remembering your preferences and repeat visits. By clicking “Accept All”, you consent to the use of ALL the cookies. However, you may visit "Cookie Settings" to provide a controlled consent.
This website uses cookies to improve your experience while you navigate through the website. Out of these, the cookies that are categorized as necessary are stored on your browser as they are essential for the working of basic functionalities of the website. We also use third-party cookies that help us analyze and understand how you use this website. These cookies will be stored in your browser only with your consent. You also have the option to opt-out of these cookies. But opting out of some of these cookies may affect your browsing experience.
Necessary cookies are absolutely essential for the website to function properly. These cookies ensure basic functionalities and security features of the website, anonymously.
Cookie
Duration
Description
cookielawinfo-checkbox-analytics
11 months
This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Analytics".
cookielawinfo-checkbox-functional
11 months
The cookie is set by GDPR cookie consent to record the user consent for the cookies in the category "Functional".
cookielawinfo-checkbox-necessary
11 months
This cookie is set by GDPR Cookie Consent plugin. The cookies is used to store the user consent for the cookies in the category "Necessary".
cookielawinfo-checkbox-others
11 months
This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Other.
cookielawinfo-checkbox-performance
11 months
This cookie is set by GDPR Cookie Consent plugin. The cookie is used to store the user consent for the cookies in the category "Performance".
viewed_cookie_policy
11 months
The cookie is set by the GDPR Cookie Consent plugin and is used to store whether or not user has consented to the use of cookies. It does not store any personal data.
Functional cookies help to perform certain functionalities like sharing the content of the website on social media platforms, collect feedbacks, and other third-party features.
Performance cookies are used to understand and analyze the key performance indexes of the website which helps in delivering a better user experience for the visitors.
Analytical cookies are used to understand how visitors interact with the website. These cookies help provide information on metrics the number of visitors, bounce rate, traffic source, etc.
Advertisement cookies are used to provide visitors with relevant ads and marketing campaigns. These cookies track visitors across websites and collect information to provide customized ads.
