Redmond, Washington, United States

Overview Are you interested in developing and optimizing deep learning systems? Are you interested in designing novel technology to accelerate their training and serving for cutting edge models and applications? Do you want to scale large Artificial Intelligence models to their limits on massive supercomputers? Are you interested in being part of an exciting open-source library for deep learning systems? The DeepSpeed team is hiring!

Microsoft's DeepSpeed is an open-source library built on the PyTorch (machine learning framework) ecosystem that combines numerous research innovations and technology advancements to make deep learning efficient and easier to use. DeepSpeed can parallelize across thousands of GPUs and train models with trillions of parameters. Our OSS (Open Source Software) has powered many advanced models like MT-530B and BLOOM, and it supports unprecedented scale and speed for both training and inference.

The DeepSpeed team is also part of the larger Microsoft AI at Scale initiative, which is pioneering the next-generation AI capabilities that are scaled across the company’s products and AI platforms.

The DeepSpeed team is looking for a Senior Researcher in Redmond, WA with passion for innovations and for building high-quality systems that will make significant impact inside and outside of Microsoft. Our team is highly collaborative, innovative, and end-user obsessed. We are looking for candidates with systems skills and passionate about driving innovations to improve the efficiency and effectiveness of deep learning systems. We value creativity, agility, accountability, and a desire to learn new technologies.

Microsoft’s mission is to empower every person and every organization on the planet to achieve more. As employees we come together with a growth mindset, innovate to empower others, and collaborate to realize our shared goals. Each day we build on our values of respect, integrity, and accountability to create a culture of inclusion where everyone can thrive at work and beyond.

Responsibilities Excels in one or more subareas and gains expertise in a broad area of research. Identifies and articulates problems in an area of research that are academically novel and may directly or indirectly impact business opportunities. Collaborates with other relevant researchers or research groups to contribute to or advance a research agenda. Researches and develops an understanding of the state-of-the-art insights, tools, technologies, or methods being used in the research community. Expands collaborative relationships with relevant product and business groups inside or outside of Microsoft and provides expertise or technology to them.

Seattle, WA; Costa Mesa, CA; Boston, MA

Anduril Industries is a defense technology company with a mission to transform U.S. and allied military capabilities with advanced technology. By bringing the expertise, technology, and business model of the 21st century’s most innovative companies to the defense industry, Anduril is changing how military systems are designed, built and sold. Anduril’s family of systems is powered by Lattice OS, an AI-powered operating system that turns thousands of data streams into a realtime, 3D command and control center. As the world enters an era of strategic competition, Anduril is committed to bringing cutting-edge autonomy, AI, computer vision, sensor fusion, and networking technology to the military in months, not years.

THE ROLE We build Lattice, the foundation for everything we do as a defense technology company. Our engineers are talented and hard-working, motivated to see their work rapidly deployed on the front lines. Our team is not just building an experiment in waiting, we deploy what we build on the Southern border, Iraq, Ukraine and more.

We have open roles across Platform Engineering, ranging from core infrastructure to distributed systems, web development, networking and more. We hire self-motivated people, those who hold a higher bar for themselves than anyone else could hold for them. If you love building infrastructure, platform services, or just working in high performing engineering cultures we invite you to apply!

YOU SHOULD APPLY IF YOU: -At least 3+ years working with a variety of programming languages such as Rust, Go, C++, Java, Python, JavaScript/TypeScript, etc. -Have experience working with customers to deliver novel software capabilities -Want to work on building and integrating model/software/hardware-in-the-loop components by leveraging first and third party technologies (related to simulation, data management, compute infrastructure, networking, and more). -Love building platform and infrastructure tooling that enables other software engineers to scale their output -Enjoy collaborating with team members and partners in the autonomy domain, and building technologies and processes which enable users to safely and rapidly develop and deploy autonomous systems at scale. -Are a U.S. Person because of required access to U.S. export controlled information

Note: The above bullet points describe the ideal candidate. None of us matched all of these at once when we first joined Anduril. We encourage you to apply even if you believe you meet only part of our wish list.

NICE TO HAVE -You've built or invented something: an app, a website, game, startup -Previous experience working in an engineering setting: a startup (or startup-like environment), engineering school, etc. If you've succeeded in a low structure, high autonomy environment you'll succeed here! -Professional software development lifecycle experience using tools such as version control, CICD systems, etc. -A deep, demonstrated understanding of how computers and networks work, from a single desktop to a multi-cluster cloud node -Experience building scalable backend software systems with various data storage and processing requirements -Experience with industry standard cloud platforms (AWS, Azure), CI/CD tools, and software infrastructure fundamentals (networking, security, distributed systems) -Ability to quickly understand and navigate complex systems and established code bases -Experience implementing robot or autonomous vehicle testing frameworks in a software-in-the-loop or hardware-in-the-loop (HITL) environment -Experience with modern build and deployment tooling (e.g. NixOS, Terraform) -Experience designing complex software systems, and iterating upon designs via a technical design review process -Familiarity with industry standard monitoring, logging, and data management tools and best practices -A bias towards rapid delivery and iteration

We are seeking a highly motivated candidate for a fully funded postdoctoral researcher position to work in 3D computer graphics and 3D computer vision.

The successful candidate will join the 3D Graphics and Vision research group led by Prof. Binh-Son Hua at the School of Computer Science and Statistics, Trinity College Dublin, Ireland to work on topics related to generative AI in the 3D domain. The School of Computer Science and Statistics at Trinity College Dublin is a collegiate, friendly, and research-intensive centre for academic study and research excellence. The School has been ranked #1 in Ireland, top 25 in Europe, and top 100 Worldwide (QS Subject Rankings 2018, 2019, 2020, 2021).

The postdoctoral researcher is expected to conduct fundamental research and publish in top-tier computer vision and computer graphics conferences (CVPR, ECCV, ICCV, SIGGRAPH) and journals (TPAMI, IJCV). Other responsibilities include supporting graduate or undergraduate students with technical guidance and engagement in other research activities such as paper reviews, reading group, workshop organization, etc.

The start date of the position is August 01, 2024. Contract duration is 1 year with the option of renewing for a second year. The successful candidate will require the following skills and knowledge: • PhD in Computer Science or related fields; • Strong tracked records in 3D computer graphics, 3D computer vision; • Hands-on experience in training deep models and generative models is required; • Hands-on experience and relevant skills in computer graphics and computer vision application development such as OpenGL, OpenCV, CUDA, Blender is desirable; • Strong programming skills in C++, Python. Capability in implementing systems from research papers and open-source software. • Additional background in math, statistics, or physics is an advantage.

Applicants should provide the following information: • A comprehensive CV including a full list of publications; • The name and contact details of two referees. One of the referees should be the applicant’s PhD supervisor; • Two representative papers by the applicant. Interested candidates should email their applications to Binh-Son Hua (https://sonhua.github.io) directly. Applications will be reviewed on a rolling basis until the position has been filled.

Open to Seattle, WA; Costa Mesa, CA; or Washington, DC

Anduril Industries is a defense technology company with a mission to transform U.S. and allied military capabilities with advanced technology. By bringing the expertise, technology, and business model of the 21st century’s most innovative companies to the defense industry, Anduril is changing how military systems are designed, built and sold. Anduril’s family of systems is powered by Lattice OS, an AI-powered operating system that turns thousands of data streams into a realtime, 3D command and control center. As the world enters an era of strategic competition, Anduril is committed to bringing cutting-edge autonomy, AI, computer vision, sensor fusion, and networking technology to the military in months, not years.

WHY WE’RE HERE The Mission Software Engineering team builds, deploys, integrates, extends, and scales Anduril's software to deliver mission-critical capabilities to our customers. As the software engineers closest to Anduril customers and end-users, Mission Software Engineers solve technical challenges of operational scenarios while owning the end-to-end delivery of winning capabilities such as Counter Intrusion, Joint All Domain Command & Control, and Counter-Unmanned Aircraft Systems.

As a Mission Software Engineer, you will solve a wide variety of problems involving networking, autonomy, systems integration, robotics, and more, while making pragmatic engineering tradeoffs along the way. Your efforts will ensure that Anduril products seamlessly work together to achieve a variety of critical outcomes. Above all, Mission Software Engineers are driven by a “Whatever It Takes” mindset—executing in an expedient, scalable, and pragmatic way while keeping the mission top-of-mind and making sound engineering decisions to deliver successful outcomes correctly, on-time, and with high quality.

WHAT YOU’LL DO -Own the software solutions that are deployed to customers -Write code to improve products and scale the mission capability to more customers -Collaborate across multiple teams to plan, build, and test complex functionality -Create and analyze metrics that are leveraged for debugging and monitoring -Triage issues, root cause failures, and coordinate next-steps -Partner with end-users to turn needs into features while balancing user experience with engineering constraints -Travel up to 30% of time to build, test, and deploy capabilities in the real world

CORE REQUIREMENTS -Strong engineering background from industry or school, ideally in areas/fields such as Computer Science, Software Engineering, Mathematics, or Physics. -At least 2-5+ years working with a variety of programming languages such as Java, Python, Rust, Go, JavaScript, etc. (We encourage all levels to apply) -Experience building software solutions involving significant amounts of data processing and analysis -Ability to quickly understand and navigate complex systems and established code bases -A desire to work on critical software that has a real-world impact -Must be eligible to obtain and maintain a U.S. TS clearance

Desired Requirements -Strong background with focus in Physics, Mathematics, and/or Motion Planning to inform modeling & simulation (M&S) and physical systems -Developing and testing multi-agent autonomous systems and deploying in real-world environments. -Feature and algorithm development with an understanding of behavior trees. -Developing software/hardware for flight systems and safety critical functionality. -Distributed communication networks and message standards -Knowledge of military systems and operational tactics

WHAT WE VALUE IN MISSION SOFTWARE Customer Facing - Mission Software Engineers are the software engineers closest to Anduril customers, end-users, and the technical challenges of operational scenarios. Mission First - Above all, MSEs execute their mission in an expedient, scalable, and pragmatic way. They keep the mission top-

Vancouver, British Columbia, Canada

Overview Microsoft Research (MSR), a leading industrial research laboratory comprised of over 1,000 computer scientists working across the United States, United Kingdom, China, India, Canada, and the Netherlands.

We are currently seeking a Senior Researcher in the area of  Artificial Specialized Intelligence located in Vancouver, British Columbia, with a keen interest in developing cutting-edge large foundation models and post-training techniques for different domains and scenarios. This is an opportunity to drive an ambitious research agenda while collaborating with diverse teams to push for novel applications of those areas.  
  Over the past 30 years, our scientists have not only conducted world-class computer science research but also integrated advanced technologies into our products and services, positively impacting millions of lives and propelling Microsoft to the forefront of digital transformation.

Responsibilities Conduct cutting-edge research in large foundation models, focusing on applying large foundation models in specific domain. Collaborate with cross-functional teams to integrate solutions into Artificial Intelligence (AI) -driven system. Develop and maintain research prototypes and software tools, ensuring that they are well-documented and adhere to best practices in software development. Publish research findings in top-tier conferences and journals and present your work at industry events. Collaborate with other AI researchers and engineers, sharing knowledge and expertise to foster a culture of innovation and continuous learning within the team.

Zoox is looking for a software engineer to join our Perception team and help us build novel architectures for classifying and understanding the complex and dynamic environments in our cities. In this role, you will have access to the best sensor data in the world and an incredible infrastructure for testing and validating your algorithms. We are creating new algorithms for segmentation, tracking, classification, and high-level scene understanding, and you could work on any (or all!) of these components.

We're looking for engineers with advanced degrees and experience building perception pipelines that work with real data in rapidly changing and uncertain environments.

Location Seattle, WA

Description Amazon Advertising is one of Amazon's fastest growing businesses. Amazon's advertising portfolio helps merchants, retail vendors, and brand owners succeed via native advertising, which grows incremental sales of their products sold through Amazon. The primary goals are to help shoppers discover new products they love, be the most efficient way for advertisers to meet their business objectives, and build a sustainable business that continuously innovates on behalf of customers. Our products and solutions are strategically important to enable our Retail and Marketplace businesses to drive long-term growth. We deliver billions of ad impressions and millions of clicks and break fresh ground in product and technical innovations every day!

The Creative X team within Amazon Advertising time aims to democratize access to high-quality creatives (images, videos) by building AI-driven solutions for advertisers. To accomplish this, we are investing in latent-diffusion models, large language models (LLM), computer vision (CV), reinforced learning (RL), and image + video and audio synthesis. You will be part of a close-knit team of applied scientists and product managers who are highly collaborative and at the top of their respective fields.

We are looking for talented Applied Scientists who are adept at a variety of skills, especially with computer vision, latent diffusion or related foundational models that will accelerate our plans to generate high-quality creatives on behalf of advertisers. Every member of the team is expected to build customer (advertiser) facing features, contribute to the collaborative spirit within the team, publish, patent, and bring cutting edge research to raise the bar within the team.

Location San Diego

Description

Qualcomm AI Research is looking for world-class algorithm engineers in general domain machine learning, especially deep learning, generative AI, LLM, LVM. Come join a high-caliber team of engineers building advanced machine learning technology, best-in-class solutions, and user friendly model optimization tools such as Qualcomm Innovation Center’s AI Model Efficiency Toolkit (https://github.com/quic/aimet) to enable state-of-the-art networks to run on devices with limited power, memory, and computation.

Members of our team enjoy the opportunity to participate in cutting edge research while simultaneously contributing technology that will be deployed worldwide in our industry-leading devices. You will be part of a multi-disciplinary talented team working on on-device generative AI optimization. Collaborate in a cross-functional environment spanning hardware, software and systems. See your design in action on industry-leading chips embedded in the next generation of smartphones, autonomous vehicles, robotics, and IOT devices.

Minimum Qualifications: • Bachelor's degree in Computer Science, Engineering, Information Systems, or related field and 4+ years of Hardware Engineering, Software Engineering, Systems Engineering, or related work experience. OR Master's degree in Computer Science, Engineering, Information Systems, or related field and 3+ years of Hardware Engineering, Software Engineering, Systems Engineering, or related work experience. OR PhD in Computer Science, Engineering, Information Systems, or related field and 2+ years of Hardware Engineering, Software Engineering, Systems Engineering, or related work experience.The R&D work responsibility for this position focuses on the following: Algorithms research and development in the area of Generative AI, LVM, LLM, Multi-modality Efficient inference algorithms research and development, e.g. batching, KV caching, efficient attentions, long context, speculative decoding Advanced quantization algorithms research and development for complex generative models, e.g., gradient/non-gradient based optimization, equivalent/non-equivalent transformation, automatic mixed precision, hardware in loop Model compression, lossy or lossless, structural and neural search Optimization based learning and learning based optimization Generative AI system prototyping Apply solutions toward system innovations for model efficiency advancement on device as well as in the cloud Python, Pytorch programmer Preferred Qualifications: Master's degree in Computer Science, Engineering, Information Systems, or related field. PHD's degree is preferred. 2+ years of experience with Machine Learning algorithms or systems engineering or related work experience

Location Mountain View, CA

Gatik is thrilled to be at CVPR! Come meet our team at booth 1831 to talk about how you could make an impact at the autonomous middle mile logistics company redefining the transportation landscape.

Who we are: Gatik, the leader in autonomous middle mile logistics, delivers goods safely and efficiently using its fleet of light & medium-duty trucks. The company focuses on short-haul, B2B logistics for Fortune 500 customers including Kroger, Walmart, Tyson Foods, Loblaw, Pitney Bowes, Georgia-Pacific, and KBX; enabling them to optimize their hub-and-spoke supply chain operations, enhance service levels and product flow across multiple locations while reducing labor costs and meeting an unprecedented expectation for faster deliveries. Gatik’s Class 3-7 autonomous box trucks are commercially deployed in multiple markets including Texas, Arkansas, and Ontario, Canada.

About the role:

We're currently looking for a tech lead with specialized skills in LiDAR, camera, and radar perception technologies to enhance our autonomous driving systems' ability to understand and interact with complex environments. In this pivotal role, you'll be instrumental in designing and refining the ML algorithms that enable our trucks to safely navigate and operate in complex, dynamic environments. You will collaborate with a team of experts in AI, robotics, and software engineering to push the boundaries of what's possible in autonomous trucking.

What you'll do: - Design and implement cutting-edge perception algorithms for autonomous vehicles, focusing on areas such as sensor fusion, 3D object detection, segmentation, and tracking in complex dynamic environments - Design and implement ML models for real-time perception tasks, leveraging deep neural networks to enhance the perception capabilities of self-driving trucks - Lead initiatives to collect, augment, and utilize large-scale datasets for training and validating perception models under various driving conditions - Develop robust testing and validation frameworks to ensure the reliability and safety of the perception systems across diverse scenarios and edge cases - Conduct field tests and simulations to validate and refine perception algorithms, ensuring robust performance in real-world trucking routes and conditions - Work closely with the data engineering team to build and maintain large-scale datasets for training and evaluating perception models, including the development of data augmentation techniques

**Please click on the Apply link below to see the full job description and apply.

Location San Francisco, CA

Description Amazon Music is an immersive audio entertainment service that deepens connections between fans, artists, and creators. From personalized music playlists to exclusive podcasts, concert livestreams to artist merch, Amazon Music is innovating at some of the most exciting intersections of music and culture. We offer experiences that serve all listeners with our different tiers of service: Prime members get access to all the music in shuffle mode, and top ad-free podcasts, included with their membership; customers can upgrade to Amazon Music Unlimited for unlimited, on-demand access to 100 million songs, including millions in HD, Ultra HD, and spatial audio; and anyone can listen for free by downloading the Amazon Music app or via Alexa-enabled devices. Join us for the opportunity to influence how Amazon Music engages fans, artists, and creators on a global scale.

You will be managing a team within the Music Machine Learning and Personalization organization that is responsible for developing, training, serving and iterating on models used for personalized candidate generation for both Music and Podcasts.

Geomagical Labs is a 3D R&D lab, in partnership with IKEA. We create magical mixed-reality experiences for hundreds of millions of users, using computer vision, neural networks, graphics, and computational photography. Last year we launched IKEA Kreativ, and we’re excited for what’s next! We have an opening in our lab for a senior computer vision researcher, with 3D Reconstruction and Deep Learning expertise, to develop and improve the underlying algorithms powering our consumer products. We are looking for highly-motivated, creative, applied researchers with entrepreneurial drive, that are excited about building novel technologies and shipping them all the way to the hands of millions of customers!

Requirements: Ph.D. and 2+ years of experience, or Master's and 6+ years of experience, focused on 3D Computer Vision and Deep Learning. Experience in classical methods for 3D Reconstruction: SfM/SLAM, Multi-view Stereo, RGB-D Fusion. Experience in using Deep Learning for 3D Reconstruction and/or Scene Understanding, having worked in any of: Depth Estimation, Room Layout Estimation, NeRFs, Inverse Rendering, 3D Scene Understanding. Familiarity with Computer Graphics and Computational Photography. Expertise in ML frameworks and libraries, e.g. PyTorch. Highly productive in Python. Ability to architect and implement complex systems at the micro and macro level. Entrepreneurial: Adventurous, self-driven, comfortable under uncertainty, with a desire to make systems work end-to-end. Innovative; with a track record of patents and/or first-authored publications at leading workshops or conferences such as CVPR, ECCV/ICCV, SIGGRAPH, ISMAR, NeurIPS, ICLR etc. Experience in developing technologies that got integrated into products, as well as post-launch performance tracking and shipping improvements. [Bonus] Comfortable with C++.

Benefits: Join a mission-driven R&D lab, strategically backed by an influential global brand. Work in a dynamic team of computer vision, AI, computational photography, AR, graphics, and design professionals, and successful serial entrepreneurs. Opportunity to publish novel and relevant research. Fully remote work available to people living in the USA or Canada. Headquartered in downtown Palo Alto, California --- an easy walk from restaurants, coffee shops and Caltrain commuter rail. The USA base salary for this full-time position ranges from $180,000 to $250,000 determined by location, role, skill, and experience level. Geomagical Labs offers a comprehensive set of benefits, and for qualifying roles, substantial incentive grants, vesting annually.

Location Multiple Locations

Description

Qualcomm's Multimedia R&D and Standards Group is seeking candidates for Video Compression Research Engineer positions. You will be part of world-renowned team of video compression experts. The team develops algorithms, hardware architectures, and systems for state-of-the-art applications of classical and machine learning methods in video compression, video processing, point cloud coding and processing, AR/VR and computer vision use cases. The successful candidate for this position will be a highly self-directed individual with strong creative and analytic skills and a passion for video compression technology. You will work on, but not be limited to, developing new applications of classical and machine learning methods in video compression improving state-of-the-art video codecs.

We are considering candidates with various levels of experience. We are flexible on location and open to hiring anywhere, preferred locations are USA, Germany and Taiwan.

Responsibilities: Contribute to the conception, development, implementation, and optimization of new algorithms extending existing techniques and systems allowing improved video compression. Initiate ideas, design and implement algorithms for superior hardware encoder performance, including perceptually based bit allocation. Develop new algorithms for deep learning-based video compression solutions. Represent Qualcomm in the related standardization forums: JVET, MPEG Video, and ITU-T/VCEG. Document and present new algorithms and implementations in various forms, including standards contributions, patent applications, conference and journal publications, presentations, etc. Ideal candidate would have the skills/experience below: Expert knowledge of the theory, algorithms, and techniques used in video and image coding. Knowledge and experience of video codecs and their test models, such as ECM, VVC, HEVC and AV1. Experience with deep learning structures CNN, RNN, autoencoder etc. and frameworks like TensorFlow/PyTorch. Track record of successful research accomplishments demonstrated through published papers, and/or patent applications in the fields of video coding or video processing. Solid programming and debugging skills in C/C++. Strong written and verbal English communication skills, great work ethic, and ability to work in a team environment to accomplish common goals. PhD or Masters degree in Electrical Engineering, Computer Science, Physics, Mathematics or similar field, or equivalent practical experience.

Qualifications: PhD or Masters degree in Electrical Engineering, Computer Science, Physics, Mathematics, or similar fields. 1+ years of experience with programming language such as C, C++, MATLAB, etc.

Location Multiple Locations

Description Today, more intelligence is moving to end devices, and mobile is becoming the pervasive AI platform. Building on the smartphone foundation and the scale of mobile, Qualcomm envisions making AI ubiquitous—expanding beyond mobile and powering other end devices, machines, vehicles, and things. We are inventing, developing, and commercializing power-efficient on-device AI, edge cloud AI, and 5G to make this a reality.

Job Purpose & Responsibilities As a member of Qualcomm’s ML Systems Team, you will participate in two activities: Development and evolution of ML/AI compilers (production and exploratory versions) for efficient mappings of ML/AI algorithms on existing and future HW Analysis of ML/AI algorithms and workloads to drive future features in Qualcomm’s ML HW/SW offerings

Key Responsibilities: Contributing to the development and evolution of ML/AI compilers within Qualcomm Defining and implementing algorithms for mapping ML/AI workloads to Qualcomm HW Understanding trends in ML network design, through customer engagements and latest academic research, and how this affects both SW and HW design Creation of performance-driven simulation components (using C++, Python) for analysis and design of high-performance HW/SW algorithms on future SoCs Exploration and analysis of performance/area/power trade-offs for future HW and SW ML algorithms Pre-Silicon prediction of performance for various ML algorithms Running, debugging and analyzing performance simulations to suggest enhancements to Qualcomm hardware and software to tackle compute and system memory-related bottlenecks · Successful applications will work in cross-site, cross-functional teams.

Requirements: Demonstrated ability to learn, think and adapt in fast changing environment Detail-oriented with strong problem-solving, analytical and debugging skills Strong communication skills (written and verbal) Strong background in algorithm development and performance analysis is essential The following experiences would be significant assets: Strong object-oriented design principles Strong knowledge of C++ Strong knowledge of Python Experience in compiler design and development Knowledge of network model formats/platforms (eg. Pytorch, Tensorflow, ONNX) is an asset. On-silicon debug skills of high-performance compute algorithms · Knowledge of algorithms and data structures Knowledge of software development processes (revision control, CD/CI, etc.) · Familiarity with tools such as git, Jenkins, Docker, clang/MSVC Knowledge of computer architecture, digital circuits and event-driven transactional models/simulators

We are looking for a Research Engineer, with passion for working on cutting edge problems that can help us create highly realistic, emotional and life-like synthetic humans through text-to-video.

Our aim is to make video content creation available for all - not only to studio production!

🧑🏼‍🔬 You will be someone who loves to code and build working systems. You are used to working in a fast-paced start-up environment. You will have experience with the software development life cycle, from ideation through implementation, to testing and release. You will also have extensive knowledge and experience in Computer Vision domain. You will also have experience within Generative AI space (GANs, Diffusion models and the like!).

👩‍💼 You will join a group of more than 50 Engineers in the R&D department and will have the opportunity to collaborate with multiple research teams across diverse areas, our R&D research is guided by our co-founders - Prof. Lourdes Agapito and Prof. Matthias Niessner and director of Science Prof. Vittorio Ferrari.

If you know and love DALL.E, MUSE, IMAGEN, MAKE-A-VIDEO, STABLE DIFFUSION and more - and you love large data, large compute and writing clean code, then we would love to talk to you.

Redmond, Washington, United States

Overview We are seeking a Principal Research Engineer to join our organization and help improve steerability and control Large Language Models (LLMs) and other AI systems. Our team currently develops Guidance, a fully open-source project that enables developers to control language models more precisely and efficiently with constrained decoding.

As a Principal Research Engineer, you will play a crucial role in advancing the frontier of constrained decoding and imagining new application programming interface (APIs) for language models. If you’re excited about links between formal grammars and generative AI, deeply understanding and optimizing LLM inference, enabling more responsible AI without finetuning and RLHF, and/or exploring fundamental changes to the “text-in, text-out” API, we’d love to hear from you. Our team offers a vibrant environment for cutting-edge, multidisciplinary research. We have a long track record of open-source code and open publication policies, and you’ll have the opportunity to collaborate with world-leading experts across Microsoft and top academic institutions across the world.

Microsoft’s mission is to empower every person and every organization on the planet to achieve more. As employees we come together with a growth mindset, innovate to empower others, and collaborate to realize our shared goals. Each day we build on our values of respect, integrity, and accountability to create a culture of inclusion where everyone can thrive at work and beyond. In alignment with our Microsoft values, we are committed to cultivating an inclusive work environment for all employees to positively impact our culture every day.

Responsibilities Develop and implement new constrained decoding research techniques for increasing LLM inference quality and/or efficiency. Example areas of interest include speculative execution, new decoding strategies (e.g. extensions to beam search), “classifier in the loop” decoding for responsible AI, improving AI planning, and explorations of attention-masking based constraints. Re-imagine the use and construction of context-free grammars (CFG) and beyond to fit Generative AI. Examples of improvements here include better tools for constructing formal grammars, extensions to Earley parsing, and efficient batch processing for constrained generation. Consideration of how these techniques are presented to developers – who may not be well versed in grammars and constrained generation -- in an intuitive, idiomatic programming syntax is also top of mind. Design principled evaluation frameworks and benchmarks for measuring the effects of constrained decoding on a model. Some areas of interest to study carefully include efficiency (token throughput and latency), generation quality, and impacts of constrained decoding on AI safety. Publish your research in top AI conferences and contribute your research advances to the guidance open-source project. Other

Embody our Culture and Values