Software Engineer, Dialpad
Large-Scale LLM Training
CA$156.6-182.5k
AI-powered cloud communications platform
Open for applications
AI-powered cloud communications platform
1001+ employees
Open for applications
CA$156.6-182.5k
1001+ employees
Company mission
To completely transform the way the world works together.
Role
Who you are
- Master’s or PhD degree in Computer Science, Machine Learning, or a related field
- 3+ years of experience in distributed systems or machine learning infrastructure
- Strong proficiency in Python and experience with distributed ML frameworks (e.g., PyTorch, TensorFlow, Horovod, or Ray)
- Understanding of networking protocols and distributed communication libraries (e.g., NCCL, gRPC)
- Hands-on experience with cloud platforms (AWS, GCP, Azure) and cluster orchestration tools (Kubernetes, Slurm)
- Proven ability to debug and resolve issues in large-scale distributed systems
- Familiarity with fault tolerance, caching strategies, and scalable logging systems
- Excellent problem-solving and communication skills
What the job involves
- As a Software Engineer, Large-Scale LLM Training at Dialpad, you'll be an integral part of our NLP team, working on our in-house LLM, DialpadGPT and other LLMs and addressing the unique challenges of training large models on big data across third party and on-prem distributed infrastructure.
- You will work on maintaining a robust multi-node training infrastructure and clusters, optimizing distributed training workflows including data loading and logging, debugging complex system errors or hardware failures, and ensuring efficient utilization of resources
- If you're passionate about large scale machine learning, generative AI, and contributing to a team that's changing the face of business communications, you'll find yourself right at home with us
- This position reports to the Manager of the NLP team and has the opportunity to be based in our office in Vancouver, BC or Kitchener, ON
- Enhance LLM training efficiency by optimizing scripts and architectures, leveraging CUDA and advanced GPU acceleration techniques to improve performance and reduce training time
- Optimize preprocessing pipelines by resolving timeout issues and implementing caching strategies
- Ensure reliable distributed networking by addressing connection failures and implementing monitoring systems
- Minimize downtime on rental machines through efficient recovery workflows and off-hour error mitigation
- Streamline debugging of distributed systems, addressing node failures and network partitions
- Develop scalable logging frameworks for efficient error detection and resolution
- Optimize system scalability by balancing workloads and reducing communication overhead
- Design fault-tolerant systems to handle node crashes and ensure seamless job restarts
- Monitor system health and implement recovery strategies for unresponsive nodes
- Manage massive data sets for large-scale model training workflows
- Develop real-time monitoring frameworks for training processes to detect and alert issues promptly
Our take
Dialpad offers an AI-driven cloud communications platform for business teams. It was founded to streamline communications across entire businesses with a fast set-up and hassle-free deployment.
As well as hosting in-house video, meet and text communications with automated note taking and transcription, the platform also features AI-powered customer service, sales and voice intelligence insights. Dialpad has impressed customers such as Asana, Stripe, TED and Uber with its smooth integration capabilities with co-working platforms like Slack, Google Workspace, Salesforce CRM and Microsoft Teams.
The Unicorn company's funding has allowed it to heavily invest in the continued R&D of AI technologies to deliver a number of industry-first features. The company has also expanded globally with the opening of offices in Europe and Asia and plans to further extend its global footprint with additional investments internationally.
Freddie
Company Specialist at Welcome to the Jungle
Insights
8% employee growth in 12 months
Company
Funding (last 2 of 9 rounds)
Dec 2022
$50m
LATE VC
Dec 2021
$170m
LATE VC
Total funding: $450m
Company benefits
- Company stock options
- Continued education stipend
- Cell phone, home internet, and gym membership stipend
- Work from home opportunities
Company HQ
San Ramon, CA
Leadership
Craig Walker
(CEO)Having studied Law at UCB, Craig worked as an Attorney for Brobeck Phleger & Harrison, before moving to General Counsel at TeleSoft Partners. He spent almost 4 years at Google Ventures before founding Dialpad in January 2011.
Salary benchmarks
We don't have enough data yet to provide salary benchmarks for this role.
Submit your salary to help other candidates with crowdsourced salary estimates.
Share this job
View 40 more jobs at Dialpad