Year-round
Research Scientist Intern
Audio, PhD
Posted on 11/15/2025
Meta
10,001+ employees
Global social networks and advertising platform
Compensation Overview
$44.13 - $70.00/hr
Company Historically Provides H1B Sponsorship
Redmond, WA, USA
Hybrid
US Top Secret Clearance Required
 AI & Machine Learning (2) |
|---|
| , |
- Currently is in the process of obtaining a PhD in the field of AI/ML and Audio/Speech Signal Processing
- Must obtain work authorization in the country of employment at the time of hire, and maintain ongoing work authorization during employment
- Research experience in one or more of these areas: machine learning, deep learning, audio/speech processing, ML model compression, or related fields
- Experience building novel computational models in audio or speech application domains using machine learning and/or signal processing
- Experience with Python/shell scripts/Matlab/C/C++ or similar
- Experience working with machine learning libraries such as Pytorch and Tensorflow
- Research, model, design, develop and test novel audio and speech processing algorithms using a combination of machine learning and signal processing to tackle unsolved real-world problems and push the state of the art in audio and advance AR/VR experiences
- Lead and contribute to cutting-edge AI model research that leads to publications on top-tier Audio/ML conferences
- Independently design and implement algorithms, train advanced AI models on large datasets, and evaluate their performance
- Develop novel deep learning techniques, to achieve state-of-the-art accuracy within the constraints of on-device and real-time execution
- Collaborate with other research scientists and software engineers to develop innovative deep learning techniques for audio use-cases
- Communicate the experimental results and the recommendations clearly, both within the group as well as to the cross-functional groups
- Intent to return to degree-program after the completion of the internship
- Proven track record of publications at ICASSP, Interspeech, WASPAA, IWAENC, IEEE TASLP, Neurips or similar
- Demonstrated software engineer experience via an internship, work experience, coding competitions, or widely used contributions in open source repositories (e.g. Github)
- Experience solving complex problems and comparing alternative solutions, trade offs, and diverse points of view to determine a path forward
- Experience working and communicating cross functionally in a team environment
- Strong background in one or more areas in Digital Signal Processing and Machine Learning such as: 1) Experience with research and development of real-time audio and speech digital signal processing solutions from concept to shipping on resource-limited devices, 2) Experience with developing scalable machine learning models that can evolve with newer device generations with minimal additional data, 3) Experience with techniques for model compression and/or deploying ML models on MIPS/Memory/Power constrained DSP devices, 4) Experience with large scale model training, implementing algorithms, and evaluating performance with objective or subjective audio metrics
- Deep subject matter expertise in one or more areas in speech/audio signal processing such as: 1) Speech Synthesis and/or Audio playback signal processing algorithms (compressors, limiters, EQs, and Automatic Gain Control), 2) Adaptive system identification, noise reduction, sound source localization, beamforming, adaptive filters (echo/feedback cancellation), 3) Spatial audio and room acoustics
Meta Platforms Inc. runs a family of social apps including Facebook, Instagram, and WhatsApp to help people connect, share content, and participate in online communities. It also develops virtual reality hardware and experiences through Oculus and is exploring the metaverse. Most revenue comes from advertising, with tools that let businesses target audiences using data from its large user base, plus VR product sales and digital services. The company differentiates itself by owning multiple major social platforms, offering a scalable cross-platform ad platform, and investing in VR, AR, and AI to expand digital experiences and monetization opportunities.
Company Size
10,001+
Company Stage
IPO
Headquarters
Menlo Park, California
Founded
2004
Simplify's Take
What believers are saying
- Q1 2026 revenue surges 33% to $56.3 billion from AI-enhanced advertising.
- Wisconsin approves $1B 220MW data center online in 2027 for AI expansion.
- Analysts project stock reaching $653–$792 in five years on ad and AI scale.
What critics are saying
- EU fines Meta 1.2 billion euros in 2023 for Facebook US data transfers violating GDPR.
- European Commission fines Meta 200 million euros in April 2025 for DMA data combination breaches.
- Elsevier, Cengage sue Meta on May 5, 2026, for Llama AI copyright infringement demanding damages.
What makes Meta unique
- Meta leverages AI for precise ad placement, generating $55 billion Q1 2026 revenue.
- Meta's vast platforms including Facebook, Instagram, WhatsApp enable unmatched audience targeting.
- Meta invests in Llama AI model, powering unique visual underage detection across platforms.
Help us improve and share your feedback! Did you find this helpful?
Benefits
Stock Options
Company Equity
Mental Health Support
Flexible Work Hours
Company News
Opposition to data centres has become a significant issue ahead of the US midterm elections, with voters unseating local politicians who support them. Residents cite concerns over water pollution, noise, power demands and environmental degradation. In Missouri, four city council members lost their seats over supporting a $6 billion data centre. Similar ousters occurred in Independence, Missouri, and rural North Carolina. The backlash crosses party lines, prompting state legislatures nationwide to consider bills ranging from eliminating tax incentives to construction moratoriums. Despite generating substantial tax revenue and construction jobs, communities increasingly resist these developments. Virginia, which has the most data centres, is considering eliminating sales tax exemptions worth $1.9 billion. President Trump has acknowledged affordability concerns whilst supporting development, though his proposals lack enforcement mechanisms.
Meta is raising prices for its Quest VR headsets by $50–$100 (12–20%) from 19 April, citing a global surge in memory chip prices affecting consumer electronics. However, Meta's own spending priorities have contributed to this component shortage. The company plans to spend $115–$135 billion on capital expenditures this year, up from $72 billion in 2025 and $28 billion in 2023, with most investment directed towards AI infrastructure. This includes $21 billion for data centre company CoreWeave and $10 billion for an El Paso data centre. Meta's AI spending forms part of $630 billion in industry-wide AI infrastructure investment pledged for 2026, driving up prices for RAM and GPUs. Meanwhile, Meta is reportedly planning spending cuts of up to 30% for its metaverse division, which has accumulated $73 billion in losses.
Meta and Broadcom have announced a strategic partnership under which the chipmaker will provide technology supporting Meta's training and inference accelerator chips through 2029. The deal extends Meta's custom AI chip development plans as the social media giant continues to invest in artificial intelligence infrastructure.
Meta and Broadcom have announced an extended partnership through 2029 for designing Meta's custom AI accelerators. Meta has committed to deploying one gigawatt of its training and inference accelerators under the agreement. The deal expands an existing collaboration between the two companies focused on Meta's in-house chip development. As part of the arrangement, Broadcom CEO Hock Tan has agreed to leave Meta's board of directors. Broadcom shares rose 3% in extended trading following the announcement. The partnership underscores Meta's continued investment in custom silicon to power its artificial intelligence infrastructure and reduce reliance on third-party chip suppliers.
Broadcom and Meta have announced a multi-year strategic partnership to support Meta's AI compute infrastructure through 2029. The collaboration centres on Meta Training and Inference Accelerator (MTIA) chips, with an initial deployment exceeding one gigawatt as part of a sustained multi-gigawatt rollout. The partnership will deliver what the companies call the industry's first 2nm AI compute accelerator. Broadcom will provide its XPU platform for chip co-development and advanced Ethernet technologies for networking across Meta's expanding AI compute clusters. The technology will underpin Meta's deployment of generative AI features across WhatsApp, Instagram and Threads. Meta aims to deliver what it calls "personal superintelligence" to billions of users globally. Broadcom CEO Hock Tan will transition from Meta's board to an advisory role focusing on Meta's custom silicon roadmap.