WeAreDevelopers Logo WeAreDevelopers
Sign in
For You LIVE
WeAreDevelopers LIVE • Apr 21, 2023

Optimizing your AI/ML workloads for sustainability

Sohan Maheshwar

Deployment accounts for nearly 90% of a machine learning model's energy usage. Discover how to systematically engineer your cloud architecture to balance powerful AI capabilities with planetary limitations.

Pause
Mute Enter Fullscreen
#1 about 5 min

High carbon footprint of complex machine learning models

Moving toward massive neural networks exponentially increases the energy consumption and carbon emissions required for training data.

#2 about 1 min

Reducing energy usage by migrating workloads to cloud infrastructure

Cloud data centers roughly optimize structural energy consumption compared to traditional on-premises hardware systems.

#3 about 4 min

Adopting the shared responsibility model for cloud computing sustainability

While providers optimize physical infrastructure operations, organizations must strictly govern application code efficiency and data usage to minimize environmental impact.

#4 about 4 min

Integrating sustainability goals into machine learning problem framing

Defining precise optimization targets allows teams to leverage pre-trained models and managed services instead of wasting processing resources on reinvention.

#5 about 4 min

Applying storage lifecycle policies to machine learning data

Automatically transitioning infrequently accessed records to cold storage prevents the ongoing environmental cost cycles of running high-availability server clusters.

#6 about 4 min

Optimizing log storage and file formats for reduced footprints

Implementing advanced compression algorithms and columnar file formats significantly decreases total storage volume and data transfer network overhead.

#7 about 2 min

Selecting purpose-built hardware for model training and inference

Utilizing custom silicon processors designed for intensive algorithmic processing yields inherently higher inference performance per watt than general-purpose compute instances.

#8 about 3 min

Developing machine learning models against acceptable performance thresholds

Establishing realistic architectural accuracy criteria limits the massive computing waste associated with running extended training epochs for marginal statistical gains.

#9 about 4 min

Right-sizing environments and optimizing endpoints during model deployment

Implementing asynchronous buffered application requests and multiplexed model endpoints smooths network traffic peaks entirely to eliminate heavily over-provisioned infrastructure instances.

#10 about 2 min

Negotiating impact-friendly SLAs to minimize idle capacity

Accepting acceptable minor latencies during cold failovers or batch processing runs enables absolute total infrastructure utilization on actively provisioned nodes.

#11 about 4 min

Choosing cloud data center regions powered by renewable energy

Deploying cloud infrastructure deployments closer to large renewable energy initiatives capitalizes entirely on unique regional advantages regarding green power distribution.

#12 about 4 min

Tracking normalized carbon emission metrics in cloud environments

Measuring total systemic energy consumption mathematically relative to product user growth ensures an accurate analytical reflection outlining authentic architectural operational optimizations.

#13 about 4 min

Accessing large environmental datasets to analyze global climate challenges

Processing freely available public remote sensing records intelligently empowers engineering groups everywhere to build sophisticated predictive tools forecasting complex natural global phenomena.

#14 about 6 min

Deploying machine learning models at the edge for conservation

Operating localized computer vision inferencing environments immediately on rugged hardware perimeter hardware facilitates crucial endangered ecosystem analytics regardless of internet constraints.

Matching moments

4:24 min

Balancing heavy compute demands with environmental sustainability goals

Mike Butcher Mike Butcher +3 · World Congress 2024

1:39 min

Balancing human-centric AI collaboration with environmental sustainability practices

Madalena Costa Madalena Costa · World Congress 2024

1:25 min

Addressing the sustainability and power consumption of AI

Christian Heilmann Christian Heilmann · World Congress 2026 Europe

4:03 min

Managing massive power consumption scaling in AI data centers

Stephan Gillich Stephan Gillich +3 · World Congress 2024

3:13 min

Assessing the energy demands versus sustainability benefits of computation

Alexander Gerfer Alexander Gerfer +3 · World Congress 2026 Europe

4:48 min

Establishing cross-chain accountability for sustainable machine learning usage

Alexander Gerfer Alexander Gerfer +3 · World Congress 2026 Europe

More from WeAreDevelopers LIVE

WeAreDevelopers LIVE - Is CSS Catching Up to Flash? thumbnail
1:03:33
12 days ago

WeAreDevelopers LIVE - Is CSS Catching Up to Flash?

Chris Heilmann, Daniel Cranney, Bramus Van Damme

WeAreDevelopers LIVE - Project Chopin: Make Your Team and Agents Play Along thumbnail
1:21:17
19 days ago

WeAreDevelopers LIVE - Project Chopin: Make Your Team and Agents Play Along

Chris Heilmann, Daniel Cranney, Krzysztof Cieślak

Fake or News - AI Toothbrushes, Carrier Pigeon Tech & Life on Mars - Krzysztof Cieślak thumbnail
4:23
21 days ago

Fake or News - AI Toothbrushes, Carrier Pigeon Tech & Life on Mars - Krzysztof Cieślak

Chris Heilmann, Daniel Cranney

WeAreDevelopers LIVE - Open Source Agent Toolkits thumbnail
1:00:38
26 days ago

WeAreDevelopers LIVE - Open Source Agent Toolkits

Chris Heilmann, Daniel Cranney, Eddie Jaoude

Fake or News - AI Music Detectors, Translating Ancient Languages & Tiny Heads in Teslas - Eddie Jaoude thumbnail
6:16
28 days ago

Fake or News - AI Music Detectors, Translating Ancient Languages & Tiny Heads in Teslas - Eddie Jaoude

Chris Heilmann, Daniel Cranney

Fake or News: AI Agents Hire Each Other, Spider-Man Ads in Cars & RAM Thieves - Marc Plogas thumbnail
5:46
about 1 month ago

Fake or News: AI Agents Hire Each Other, Spider-Man Ads in Cars & RAM Thieves - Marc Plogas

Chris Heilmann, Daniel Cranney, Marc Plogas

Related videos

Minimising the Carbon Footprint of Workloads thumbnail
27:46
World Congress 2024 • about 2 years ago

Minimising the Carbon Footprint of Workloads

Michael Mueller

The Sustainability Race: AI's Promises, Pitfalls and Potential thumbnail
31:18
World Congress 2026 Europe • 3 months ago

The Sustainability Race: AI's Promises, Pitfalls and Potential

Alexander Gerfer, Gülnaz Öneş, Wolfgang Oels, Awi Lifshitz

An Architect’s guide to reducing the carbon footprint of your applications thumbnail
55:18
LIVE • over 3 years ago

An Architect’s guide to reducing the carbon footprint of your applications

Ricardo Sueiras Sueiras

AI Factories at Scale thumbnail
24:26
World Congress 2024 • about 2 years ago

AI Factories at Scale

Thomas Schmidt

Green Cloud Computing thumbnail
35:20
LIVE • over 3 years ago

Green Cloud Computing

Doreen Sacker

Machine Learning for Software Developers (and Knitters) thumbnail
49:45
LIVE • over 5 years ago

Machine Learning for Software Developers (and Knitters)

Kris Howard

Related articles

From learning to earning

Jobs that call for the skills explored in this talk.