AI Innovations: The Power of Feature Platforms // MLOps Mini Summit #6
MLOps Community Mini Summit #6! We talked to Tecton's Product Manager, Mahesh Murag, Cleo's MLOps Engineer, Jose Navarro, and Fennel's Co-founder/CEO, Nikhil Garg, brought to us by @tecton.
// Abstract
Building a Unified Feature Platform for Production AI
Mahesh walks through Tecton’s journey to build a unified feature platform that powers large-scale real-time AI applications with only Python. He'll dive into how Tecton has navigated key tradeoffs like balancing developer experience with scalability and flexibility while adapting to rapidly evolving data and ML engineering best practices.
Journey of a Made-to-Measure Feature Platform at Cleo
Jose shows how the platform team at Cleo has built a production-ready Feature Platform using Feast.
DIY: How to Build a Feature Platform at Home
Nikhil decomposes a modern feature platform in key components, describes a few common options for each component, gotchas in assembling them, and some key architectural tradeoffs.
// Bio:
Mahesh Murag
Mahesh Murag is a Product Manager at Tecton focused on core product experiences, including building Tecton's unified compute engine for feature engineering at scale. Prior to joining Tecton, Mahesh was a Product Manager at Scale AI, helping customers build better models in verticals like self-driving vehicles and eCommerce. Mahesh is from the San Francisco Bay Area and has a degree in Electrical Engineering & Computer Science from the University of California, Berkeley -- he now lives in New York City.
Jose Navarro
Software engineer with +5 years of experience in startups with strong agile practices. Enthusiast for learning and proactive. Experience building, deploying, and maintaining in production both monolith and micro-services architectures. Master’s degree in Machine Learning from the University of Bristol.
Nikhil Garg
Nikhil is the cofounder/CEO of Fennel, a realtime ML Infra company. Previously, he was at Meta where he led teams building PyTorch and before that applied ML teams that handled AI/ML of many product lines across Meta. Outside of work, he is the father of an 18 month old and a major Eminem fan, in that order.
// Jobs board
https://mlops.pallet.xyz/jobs
// Related links
---------- ✌️Connect With Us ✌️------------
Join our Slack community: https://go.mlops.community/slack
Follow us on Twitter: @mlopscommunity
Sign up for the next meetup: https://go.mlops.community/register
Catch all episodes, Feature Store, Machine Learning Monitoring, and Blogs: https://mlops.community/
Connect with Demetrios on LinkedIn: / dpbrinkm
Connect with Ben on LinkedIn: / ben-epstein
Connect with Mahesh on LinkedIn: / maheshmurag
Connect with Jose on LinkedIn: / jose-navarro-2a57b612
Timestamps
[00:00] Real-time data, predictive features, and model integration
[06:34] Fast data-driven model advancement without engineering constraints
[13:49] Supports software team, data scientists' rapid iteration
[20:43] Big organizations use Python for enterprise performance
[23:26] Python enables collaboration, rapid iteration, retrieval ease
[32:23] Subscription model to avoid overdraft, credit building
[35:50]Flexibility in feature engineering platform building process
[39:18] Transformation of features for an advanced classifier, improvements
[46:24] Compiler processes definitions into actionable entities, connectors included
[50:21] Managing and storing state in a compiler
[55:53] Fennel uses rocks DB on SSD's, partitioning, replication
[01:02:55] Utilizing streaming cases for low-latency data