Staff Machine Learning Engineer, Lilt
Data Processing & Deployment
Plus equity
More information about location
2-5 days a week in office
AI solution for enterprise translation and content creation
Job no longer available
AI solution for enterprise translation and content creation
501-1000 employees
Job no longer available
Plus equity
More information about location
2-5 days a week in office
501-1000 employees
Company mission
To make the world's information accessible to everyone regardless of where they were born or which language they speak.
Role
Who you are
- Master’s degree in computer science, statistics, computational math/linguistics, machine learning, or related technical field
- 5 years of experience building large-scale Natural Language Processing systems using Machine Translation (MT) models
- Willing to accept any reasonable combination of education and experience (B.S. + 7 yrs of exp or Ph.D. + 3 yrs of exp)
- Experience using Tensorflow or PyTorch to interact with and design neural networks – 5 years
- Experience with Python – 5 years
- Experience using Nvidia NeMo to design large language models – 2 years
- Large-Scale Data Processing experience for ML Model Training – 4 years
- Experience with Production-Level Deployment of Machine Learning Models– 4 years, including at least 3 years using Kubernetes
What the job involves
- We have a strong preference for building our own backend technology, so you’ll be implementing and working with the latest natural language processing (NLP) techniques and ideas
- The Staff ML Engineer (Data Processing & Deployment) will apply knowledge of computer and information science to perform the following tasks, dividing time according to the approximate percentages set out below
- Product and Engineering (50%)
- Develop and train state-of-art production-level Machine Translation models to be used by both LILT customers and translators
- Develop and maintain a collection of services (Python, Java) to query and transform the customer data for the purpose of adaptation experiments, including but not limited to on-the-fly identification of outliers which can negatively impact fine-tuning performance
- Contribute to engineering and product planning meetings to suggest and discuss improvements to the LILT machine-learning ecosystem
- Develop automated systems to create the best possible processes for production-level deployment of Machine Learning models (Kubernetes and Helm) to provide well documented instructions for production-quality releases and A/B testing
- Identify performance bottlenecks and usability improvements in the research and development infrastructure and propose and lead improvements to replace or upgrade it with updated, more efficient, and better performing libraries and tools
- Keep up to date with libraries and technologies in the field of data caching (Redis), messaging systems (RabbitMQ, PubSub), databases (MySQL), and continuous improvement (Jenkins), to engineer the best possible product to service quality translations
- Research and Innovation (30%)
- Keep up to date on the latest research in Machine Translation, Large Language Models, and similar fields
- This involves reading and curating research papers and presenting solutions or improvements to either our product or the overall stack of Machine Translation knowledge and Large Language Models, and data preparation for human- preference alignment
- Identify and develop proof of concept solutions, such as domain adaptation, AI-driven quality measurement, for large scale Natural Language processing systems to bridge the gap between the performance of LILT and publicly or privately available Artificial Intelligence models and systems, such as GPT-4, Google Translate, and Amazon Translate
- Iterate and develop the best possible architecture for Multilingual Machine Translation and Creation models to be deployed to LILT’s production environment, while not compromising on performance
- Identify and experiment with methods to improve the processes for large-scale data processing for ML model training (reference free COMET based filtering, ROUGE, Creative Writing), to allow for faster training turnaround times while not compromising quality or performance
- Repeatedly demonstrate good judgment in driving applied research initiatives within LILT that lead to the highest possible impact for customers and internal teams
- Keep up to date with libraries and frameworks to interact with and design neural networks (Tensorflow, Pytorch, and NVIDIA NeMo)
- Submit the process and results of research projects to peer-reviewed conferences and publications (ACL, EMNLP, NAACL)
- Serve as Machine Translation Spokesperson (20%)
- Be the spokesperson for Machine Learning applied research initiatives being conducted both internally and externally LILT
- Effectively train non-technical teams (including but not limited to Production, Marketing and Sales) on complex research ideas
- Develop graphical tools and methods to enable customer-facing organizations to explain observed phenomena in Language models to customers
Application process
- 2 Positions Available
- Apply online or submit resume to Lilt, Joern Wuebker, 2200 Powell St., Ste. 900, Emeryville, CA 94608
Our take
Widely available tools like Google Translate are not fit for complex, operationally intense processes that require expert translations. The manual methods of yesterday can't scale to the volume of digital information produced today. Lilt works with companies, schools, and government agencies to provide AI-powered enterprise language translation. Its platform combines a translation management system and neural machine translation with human language experts to create a “human-in-the-loop” approach.
While companies like AI Translate offer intelligent machine translations, they still have linguists fine-tune and correct the texts afterwards. With the Lilt platform, AI and humans work together on text, with the AI making real-time sentence suggestions and the linguist reviewing them immediately. This collaboration delivers 3-5x greater efficiency than traditional human-only translation services, which has aided Lilt's substantial rounds of funding.
Clients such as Asics, Canva, and Digital Ocean use Lilt technology for urgent campaigns and large-scale localisation projects. Recently, Lilt launched its V3.5 translation models, which improve quality by 15% to 20%, significantly enhancing translation accuracy and efficiency. Additionally, Lilt appointed a new Chief Revenue Officer to drive growth and scale, underscoring its commitment to expanding its market presence and further improving its services.
Steph
Company Specialist at Welcome to the Jungle
Insights
2% employee growth in 12 months
Company
Funding (last 2 of 4 rounds)
Apr 2022
$55m
SERIES C
May 2020
$25m
SERIES B
Total funding: $91.9m
Company benefits
- Compensation: Meaningful equity, 401(k) matching, and flexible time off plus company holidays
- Medical Benefits: Employees receive coverage of medical, dental, and vision insurance, plus FSA/DFSA, HSA, and Commuter benefits. In addition, LILT pays for basic life insurance, short-term disability, and long-term disability
- Paid parental leave is provided after 6 months
- Monthly lifestyle benefit stipend via the Fringe platform to allow employees to customize benefits to their lifestyle
Company HQ
Emeryville, CA
Leadership
John DeNero
(Co-Founder & Chief Scientist)Earned a PhD in Computer Science from the University of California, Berkeley. Previously Teaching Professor at UC Berkeley for 7 years and Senior Research Scientist at Google for 4 years.
Spence Green
(CEO)Earned a Ph.D. in Computer Science from Stanford University. Previously Research Assistant - Computer Science at Stanford University for 6 years and Fellow at XSeed Capital for 1 year.
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 8 more jobs at Lilt