Overview

Preferred Networks (PFN) is looking for enthusiastic interns who can work with us remotely this summer. Students who previously applied for PFN internships are also eligible to re-apply (except those who previously interned with PFN). We welcome students who want to develop new technologies, software, and services across a wide range of computer science areas (including machine learning) with us.

We have closed the call for applications for PFN Summer Domestic Internship 2021. Thank you for many applications.

 

Index

Characteristics and Duration of a PFN Internship

PFN Intern Online Information Session

Theme List

Key Qualifications

Important Notes before You Apply

How to Apply

Selection Process

Salary and Benefits

Characteristics and Duration of PFN Internship

• Over the 1.5 month period, PFN researchers and engineers will be assigned to work with you as your mentors. You’ll have opportunities to discuss, study, and develop in your internship theme with specialists in various fields including deep learning, computer vision, robotics, bio-healthcare, reinforcement learning, and distributed processing.
• We encourage the results of the internship to be published as OSS, papers, and PFN Research & Development blogs if there are no confidentiality or rights issues.  
• PFN values the diversity of our members and the working environment. We’re looking forward to receiving applications from women who are still a minority in the IT industry. All applicants will be selected fairly.

• Start date：Tuesday, August 10th, 2021
• End date： Friday, September 17th, 2021
  • You can choose to continue to work in the week of September 21st – 30th under the same terms and conditions.
    • This year’s internship will end on September 17th in principle in consideration of the fact that many schools start their fall semester in late September. 
    • If the above case does not pertain to you and you need more time to finalize your research or want to spend more time with our staff, you can extend your internship until September 30th under the same terms and conditions. 
    • During your internship, we understand you may have to be absent for (e.g.) lab activities, attending academic conferences, and returning home for family commitments. We are flexible about your need to take days off due to these reasons.

Messages from our employees and the past interns:

Back to top

 

PFN Internship Online Information Session

The PFN Internship online information session will be held in Japanese. For details please check Japanese page.

Back to top

 

Theme List

• For the 2021 Domestic Internship at PFN, we have prepared the themes listed below for both engineering and research interns:
  • Engineering intern: emphasizes development that does not necessarily aim for research presentation (e.g. re-implementation of dissertation, development of OSS, etc.)
  • Research intern: aims to conduct research more independently and publish the results as a paper
• During or after the selection process, you will talk with PFN members to finalize your theme. You can choose up to 2 themes in the application form. 
• Any intellectual property achieved through your internship activities shall belong to PFN.  Therefore, when determining your internship theme, please ensure that you do not bring your own research theme from your university research lab or other organization you belong to.

• 3/25 Updated: Added English as a communication language in theme “JE11. Research on task mapping and architectures for next-generation deep learning accelerators.” ID has changed from JP22.

 

[Expand all] [Collapse all]

Development

[+] JE01. 3D reconstruction using machine learning

3D reconstruction techniques are growing with deep learning methods such as NeRF and IDR, as well as 3D measurement instruments like LiDAR sensors. In this project, we utilize these technologies and try to quantitatively improve existing 3D reconstruction processes.

Communication Language: English/Japanese

Must Requirements

・Familiarity with Python
・Experience in using deep learning frameworks（e.g. PyTorch, Chainer）
・Knoledge of CV or 3D

Want Requirements

・Experience in 3D reconstruction, New-view synthesis, CV.

[+] JE02. Analysis of radiological images

In this project, we will develop deep learning models to detect lesions from or perform registration of either chest X-rays, CT, or MRI scans.

Communication Language: English/Japanese

Must Requirements

・Deep knowledge of image analysis methods
・Coding skills in Python
・Experience in using deep learning frameworks（e.g., PyTorch, Chainer）

Want Requirements

・Scientific knowledge in medicine or biology (undergraduate level or above)
・Experience in R&D of image analysis or video processing
・Experience in publishing papers in the fields of medicine, biology, mathematics, or information sciences

[+] EN01. Interface Design for machine learning-based Music Creation

Preferred Networks Music Team invites internship applications on the topic of interface and software design for ML music. Possible projects include
1. Developing tools to facilitate building and testing models for music creation
2. Using ML to design new interfaces for music creation
3. Developing user-facing interfaces for in-house ML technologies.

Communication Language: English

Must Requirements

・Familiarity with Python
・Experience using PyTorch
・Knowledge of ML and deep learning
・Interest in music

Want Requirements

・Experience with music creation software such as DAWs or VSTs
・Experience with JUCE, C++, or Javascript
・Knowledge of web or app design
・Experience in real-time audio processing

[+] JE03. Development of Creator Tools

In this project, you will be working with Creative members at PFN to build tools that support artists for animation and game creation.

Communication Language: English/Japanese

Must Requirements

・Basic knowledge in computer science

Want Requirements

・Experience in using deep learning frameworks such as PyTorch
・Skills in application and web development
・Like anime/game

[+] JE04. Development of deep learning models for tabular data

In this project, we will develop and implement deep learning models using tabular data for bio and healthcare related applications. We do not require experience in data analysis for bio or healthcare data.

Communication Language: English/Japanese

Must Requirements

・Knowledge on biology or bioinformatics
・Coding skills in Python

Want Requirements

・Experience in using deep learning frameworks (e.g., PyTorch, Tensorflow, or Chainer)
・Experience in implementing deep learning models
・Experience in library development such as Python packages.

[+] JP01. コンピュータービジョンのためのwebアプリケーション開発

本テーマでは以下の開発を行います。
・動画・3Dデータなどへのアノテーションツール開発
・機械学習を使ってアノテーションを効率化するシステム開発
・データセットやモデルの管理・可視化ツール開発

Communication Language: Japanese

Must Requirements

TypeScript, Reactを用いたウェブフロントエンドの開発能力

Want Requirements

Go, Pythonのコーディング能力

[+] JP02. 深層学習・微分可能レンダラを活用した3Dスキャナ

PFNで開発している3Dスキャナの改善に取り組みます。たとえば 、以下の項目の改善を目指します。
・照明を工夫することで法線マップをキャプチャできるようにする
・ペットボトルなどの透明な物体のスキャン結果を改善する

Communication Language: Japanese

Must Requirements

・Pythonを用いたコーディング能力
・深層学習フレームワークの使用経験（PyTorch, Chainer等）

Want Requirements

三次元復元に関する研究開発経験

[+] JP03. CG生成画像を使った学習

CG生成画像を使って学習した物体検出器の精度を改善するための研究開発を行います。

Communication Language: Japanese

Must Requirements

・Familiarity with Python
・Experience using PyTorch
・Knowledge of CV and deep learning

Want Requirements

Knowledge of object detection, semantic segmentation, or instance segmentation

[+] JP04. 人物3Dモデルの自動リギング

人物3Dメッシュに対する体・手・顔などの自動リギング機能を開発します。また成果物の3Dアセットを利用し人物の姿勢推定モデルなどの学習を行います。

Communication Language: Japanese

Must Requirements

・Pythonを用いたコーディング能力
・深層学習フレームワークの使用経験（PyTorch, Chainer等）

Want Requirements

人物認識（姿勢推定、顔認識など）に取り組んだ経験

[+] JP05. 3次元物体検出手法の開発

LiDAR、画像を用いた3次元物体検出のタスクに取り組みます。

Communication Language: Japanese

Must Requirements

・画像認識または点群処理の知識
・Pythonを用いたコーディング能力
・深層学習フレームワークの使用経験（PyTorch, Chainer等）

Want Requirements

3次元物体検出に取り組んだ経験・実績（論文・学会発表があればより望ましい）

[+] JP06. 物理シミュレーションと機械学習の融合

化学工学プラント、ビルの空調、地震波解析などを対象に、物理シミュレーションと機械学習を組み合わせる手法、例えば
・機械学習によるシミュレーションの高速化
・微分可能シミュレーターの実装と機械学習モデルへの組み込み
・シミュレーション対象の構造を反映したグラフニューラルネットや物理学の知識を反映したPhysics-guidedニューラルネットを用いた予測
などのいずれかに取り組みます。

Communication Language: Japanese

Must Requirements

・計算機科学の基本的知識
・機械学習もしくはシミュレーションに関する応用的知識
・Pythonを用いたコーディング能力

Want Requirements

・深層学習フレームワークの利用経験（PyTorchを推奨）
・上記対象候補のいずれかの分野の知識経験
・共同開発の経験

[+] JP07. 音響シミュレーションの開発

実環境下での音声認識性能向上に向けて、雑音・残響・話者性などの音響シミュレーションに取り組みます。

Communication Language: Japanese

Must Requirements

・基本的なコンピュータサイエンスに関する知識
・音響音声信号処理に関する基礎的な知識

Want Requirements

音響信号処理や音声認識、音声合成に取り組んだ経験・実績（論文・学会発表があればより望ましい）

[+] JP08. 機械学習を用いた地震波解析

機械学習を用いた地震波解析に関わる技術開発に取り組みます。
・地震波解析に適した深層学習モデルの構築
・観測データとシミュレーションのギャップを埋める手法の研究開発
などの課題に取り組みます。

Communication Language: Japanese

Must Requirements

・計算機科学の基本的知識
・Pythonを用いたコーディング能力
・機械学習に関する応用的知識

Want Requirements

深層学習フレームワークの利用経験（PyTorchを推奨）

[+] JP09. 多次元時系列データを用いた異常検知手法の開発

複数のセンサーから得られる時系列データから異常検知を行う手法の開発

Communication Language: Japanese

Must Requirements

・機械学習に関する基本的知識
・異常検知アルゴリズムの開発経験

Want Requirements

実データに異常検知アルゴリズムを適用した経験

[+] JP10. 機械学習を題材としたプログラミング教育教材の開発

プログラミング教育アプリPlaygramをベースに、機械学習を題材とした教材開発を行います。

Communication Language: Japanese

Must Requirements

・日本語でのコミュニケーション能力
・プログラミングおよび機械学習の基本的な理解

Want Requirements

・Unityまたは機械学習フレームワークのいずれかにおける開発経験
・Webフレームワークに関する知識

[+] JP11. 工業プラントの制御

蒸留や化学反応を行う工業プラントの制御に関する問題として、
・人間の運転員の操作の模倣学習
・強化学習やモデル予測制御を用いた制御
・説明可能な制御手法の開発
・オフライン強化学習の問題設定での制御の評価手法
のいずれかに取り組みます。

Communication Language: Japanese

Must Requirements

・計算機科学の基本的知識
・機械学習に関する応用的知識
・Pythonを用いたコーディング能力

Want Requirements

・深層学習フレームワークの利用経験（PyTorchを推奨）
・化学工学、制御理論、数理最適化のいずれかの知識
・共同開発の経験

[+] JP12. Webサービスのための化学計算ライブラリ・システム開発

深層学習を用いた自社開発の物性値計算アルゴリズムを実サービスで利用するためのライブラリ・システムの開発を行っていただきます。

Communication Language: Japanese

Must Requirements

・Pythonを用いたコーディング能力
・深層学習フレームワークの使用経験（PyTorch, Chainer等）

Want Requirements

・Webサービスにおけるソフトウェア開発経験
・物理、化学、材料科学に関する、大学卒業程度の知識

[+] JP13. 個人に最適化した食生活のための研究開発

ライフスタイルが多様化する時代に、個人個人に最適化された食生活を提案できる情報技術の開発に取り組みます。

Communication Language: Japanese

Must Requirements

基本的なプログラミングスキル

Want Requirements

言語処理、組み合わせ最適化手法、および料理に詳しい人が望ましい

[+] JP14. 分散キャッシュの機能開発

リモートのファイルシステム上にあるISO/ZIP等のアーカイブをクライアントにマウントする際に、別のリモートホストのメモリ上へキャッシュする機能の開発に取り組みます。

Communication Language: Japanese

Must Requirements

fuse, gRPC or RDMA, C, Go or Rust, 並行制御

Want Requirements

Linuxカーネルモジュールを作れる, isofsやbtrfs に詳しい

[+] JP15. 社内向けWebサービス (flexCI) 開発

CIはアプリケーションの開発速度を大きく左右する要素であり、そのCIの機能改善を発案から設計・実装まで行います。例えば、ログのリアルタイム出力、ワークフロー機能などです。

Communication Language: Japanese

Must Requirements

Go言語によるサーバーソフトウェア開発

Want Requirements

[+] JP16. MN-Core向けコンパイラ及びライブラリ開発

MN-Coreコンパイラが必要とする高性能なカーネルの実装や、コンパイラ自体のデータ構造、プロファイラの開発などに関連する研究開発に取り組んでいただきます。

Communication Language: Japanese

Must Requirements

・計算機科学の基本的知識
・C++を用いたコーディング能力

Want Requirements

低レベル最適化の経験、深層学習向けコンパイラの知識

[+] JP17. Optuna 開発

ハイパーパラメータ自動最適化フレームワーク Optuna について
・ソフトウェア品質・設計の改善
・OSS オペレーションの改善
・実際のユースケースに向けたアルゴリズムの性能向上
に取り組みます。

Communication Language: Japanese

Must Requirements

基本的なコンピュータサイエンスに関する知識

Want Requirements

・オープンソースソフトウェアの開発・貢献またはそれに類する実績
・複数人チームでのソフトウェア開発・サポートの経験
・論文発表やコンペティションでの成果など、機械学習分野における実績
・ベイズ最適化・ニューラルアーキテクチャサーチ・自動機械学習（AutoML）についての専門知識

[+] JP18. エッジデバイスでの動作のためのフレームワーク、ライブラリ開発

エッジデバイスで低レイテンシや省電力などを考慮した深層学習モデルの推論ランタイム・ライブラリの開発を行います。

Communication Language: Japanese

Must Requirements

・C/C++, Rustを用いたコーディング能力
・コンピューターサイエンスの基礎知識

Want Requirements

・深層学習モデルの実装経験
・エッジデバイスやスマートフォンを用いた開発経験

 

Research

[+] JE05. Research and development of pathological diagnosis

In this project, we will develop deep learning models to detect lesions such as tumors from pathological images. Possible target diseases include gynecological diseases (e.g., cervical cancer) and lung cancer.

Communication Language: English/Japanese

Must Requirements

・Deep knowledge of image analysis methods
・Coding skills in Python
・Experience in using deep learning frameworks（e.g., PyTorch, Chainer）

Want Requirements

・Scientific knowledge in medicine or biology (undergraduate level or above)
・Experience in R&D of image analysis or video processing
・Experience in publishing papers in the fields of medicine, biology, mathematics, or information sciences

[+] JE06. Application of Deep Learning techniques to Animation and Game Creation

In this project, you will be working with Creative members at PFN to apply deep learning to new form of creativity in animation and game creation.

Communication Language: English/Japanese

Must Requirements

・Familiarity with Python

Want Requirements

・Experience in using deep learning frameworks such as PyTorch
・Publications in machine learning domain
・Domain knowledge in Computer Vision, Computer Graphics, Generative Model
・Skills in software development
・Like anime/game

[+] JE07. Quantitative Finance using Machine Learning

In this project, you will research applying machine learning to finance. For example, Develop a financial time-series prediction algorithm, Develop a financial trade strategy with machine learning, Develop changing point detection of the stock market.

Communication Language: English/Japanese

Must Requirements

Basic knowledge of machine learning and finance

Want Requirements

High expertise in finance-friendly fields

[+] EN02. Application of machine learning techniques to Music and Audio Creation

Preferred Networks Music Team invites research internship applications on the topic of machine learning for music creation. Working alongside our expert team of ML music researchers, you will have the opportunity to build state-of-the-art models of music generation, timber transfer, and audio synthesis.

Communication Language: English

Must Requirements

・Familiarity with Python
・Experience using PyTorch
・Knowledge of ML and deep learning
・Interest in music

Want Requirements

・Music, sound, or speech processing publications at venues such as ICASSP, INTERSPEECH, ISMIR, ICLR, ICML, NeurIPS.
・Previous experience in one or more of the following: ASR, TTS, NLP or other audio/music processing applications

[+] JE08. Resolving the gap between training and test

Collecting big data for machine learning is hard; in many real-world situations, it might cost a fortune or it might be physically impossible. This is why the world today is striving to develop methods that can train a model efficiently from small datasets, or a way to use a simulator to create a good artificial dataset. Either way, however, we must overcome the gap between the training dataset and the test dataset collected from the real environment. We are looking for diligent, excited applicants who will join us in our endeavor to develop a method for training a model that can generalize on the test environment.
Keywords: Out of distritbuion generalization, sim2real, meta-learning, control, reinforcement learning

Communication Language: English/Japanese

Must Requirements

・Record(s) of publication in top-tier conference/journal as a corresponding author
・Deep knowledge of Machine Learning

Want Requirements

・Ph.D. candidacy

・Experiences with Deep learning frameworks (preferably PyTorch)

・A research proposal that has no intersection with the current graduate thesis nor with other internship programs）

[+] JE09. Application research and development of machine learning or atomic simulation method for materials

In this project, we will develop atomic system simulators using machine learning (e.g., Beyond-DFT, electronic state prediction, NNP, coarse-grained MD), or the application of machine learning to molecular dynamics simulations and materials discovery methods (e.g., atomic structure generation, reaction pathway search, phase diagram calculations).

Communication Language: English/Japanese

Must Requirements

・Basic knowledge in computer science
・Basic knowledge or interest in physics, chemistry, or materials science

Want Requirements

・B.A. qualification or higher in physics, chemistry, or materials science
・Experience in computational chemistry. For example, quantum chemical calculation (DFT etc) or molecular dynamics simulation
・Experience in research or development on the integration of physical simulation and machine learning

[+] JE10. Research and development for semi-supervised learning for gene expression data

In this project, we will develop semi-supervised learning for gene expression data

Communication Language: English/Japanese

Must Requirements

・Knowledge in computer science (undergraduate level)
・Coding skills in Python
・Experience in using deep learning frameworks（e.g., PyTorch, Chainer）

Want Requirements

・Knowledge in biology or bioinformatics
・Experience in analysis of gene expression data (especially NGS data)

[+] [Updated] JE11. Research on task mapping and architectures for next-generation deep learning accelerators

Added English to communication languages. The ID has changed from JP22.

This research internship is about next-generation deep learning architectures and how to exploit such architectures using task mapping, routing, and scheduling algorithms. This topic can be approached from either a task mapping perspective or a computer architecture perspective.

Communication Language: English/Japanese

Must Requirements

・Solid understanding of computer architecture
・Good coding skills in any major programming language such as C++, Python, or Rust

Want Requirements

・Experience in computer architecture related research
・Experience in EDA, P&R algorithms related research

[+] JP19. 自動車の動作予測手法の研究開発

HD mapを用いた自動車の動作予測手法の研究開発のタスクに取り組みます。

Communication Language: Japanese

Must Requirements

・Pythonを用いたコーディング能力
・深層学習フレームワークの使用経験（PyTorch, Chainer等）

Want Requirements

Motion predictionに取り組んだ経験・実績（論文・学会発表があればより望ましい）

[+] JP20. 深層学習を用いた離散最適化

実世界の離散最適化問題（複雑な制約条件を持つ問題、確率変数を含む問題、時刻とともに状況が変化する問題、目的関数の明確でない問題、など）の解決ための深層学習を用いた手法の研究開発に取り組みます。

Communication Language: Japanese

Must Requirements

・コーディング能力
・計算機科学の基本知識
・機械学習の基本知識

Want Requirements

・深層学習フレームワークの利用経験
・強化学習に関する知識
・離散最適化に関する知識

[+] JP21. 創薬に関する機械学習や分子シミュレーションの開発・応用研究

創薬の分子設計に関する、
・生成モデル
・分子プロパティ予測
・結合評価
などに取り組みます。

Communication Language: Japanese

Must Requirements

・基本的なコンピュータサイエンスに関する知識
・有機化学 / 医薬化学 / 構造生物学 のいずれかについての知識

Want Requirements

・チーム開発経験/スキル
・創薬に関する生成モデル・分子プロパティ/結合予測に関して取り組んだ経験・実績（論文・学会発表があればより望ましい）

Back to top

Key Qualifications

• PFN is seeking highly motivated and skilled individuals who can develop applications, tools, etc., independently. 
• Knowledge and development experience based on the themes listed in [Theme List] will be considered in the selection process. Please check each theme for details. The common requirements are:
  • Currently enrolled in high school, technical college, university, or graduate school. Negotiable for those attending other higher education institutions
  • Fluent in Japanese or English
  • Strong communication skills
  • Can work remotely on weekdays

 

[Notes]

This year, to prevent infection, we have decided to ask interns to work from home. Due to this change, we will limit the eligibility to those who can secure a work-from-home environment to a certain degree. Here are the requirements.

• There’s no health concern about an 8h / day work as you can secure an environment with a desk, chair and light appropriate for working for a long time.
• Ensure that you can work from a physically secure location using a secure network connection.
  • When working, your PC monitor, memos containing confidential information, etc. are not visible to anyone other than your family members. You can also store them securely so that nobody other than your family members can easily access them while you’re out
  • You or your family is managing the network, and therefore nobody can easily take control of or intercept it
    • The network can be ordinary broadband. The equipment can be a normal, commercially available router
  • For example, you cannot use the following environments:
    • A room which is shared with someone other than your family members
    • Places open to the general public, e.g. lab rooms and vacant classrooms in a university, co-working spaces, cafeteria, libraries, family restaurants, fast food restaurants, hotel lobbies, parks, etc.
• During working hours, you can secure a few hours in total where you can hold a remote meeting
  • Please secure an environment where nobody except for your family members can hear your voice
    • Regarding PFN employees’ voices, please protect them by using a headphone or earphone
    • Even the room is a private room, it is not acceptable if the walls have low sound insulation
  • In case you’re living together with family members, please also make sure to get an acknowledgement from your family members about the fact that you will hold meetings and discuss where to do conduct them
• You have a network such as FTTH that doesn’t suffer from bandwidth limitation even if you use it for video and audio communication heavily
• The place of work from home is limited to Japan in consideration of related regulations and time difference.

 

If you have any questions or concerns regarding the requirements, please contact intern2021-admin@preferred.jp.

To assist in the purchase of an earphone etc., we will pay an additional 3,000 JPY in the first salary payment. If you already have an earphone, or are in an environment in which you don’t need to worry about the sound leakage, it is not required to purchase one. The assistance will be paid regardless of the necessity.

Back to top

 

Important Notes before You Apply

• Non-Japanese students who are studying at Japanese universities with a student visa must make sure to get【Permission for Other Activity / 資格外活動許可】before the start date of your PFN internship program.
• You need to let us know in advance if any administrative work is required to receive academic credit from your school. Please note that depending on the complexity of the work, PFN may not be able to accommodate your request.

Back to top

 

How to Apply

Click this Application Form to apply. 

* You need to log in with your Google account to access the application form.

* Personal information entered in the application form will be used for the purpose of intern selection.

 

[Application deadline]

April 16, 2021, 12:00 PM (Japan time,No application accepted after the deadline)

 

[Application documents]

Resume and file format are PDF only. Please submit using the application form.

 

[Notes on resume]

・Be as specific as possible in your resume so that we have all the information we need to be able to evaluate your abilities and experience.

・ The evaluators will read your resume and use it to determine whether it matches the theme you are applying for. Please be specific about your abilities and experiences related to the theme.

・Please describe your award history and research experience as comprehensively as possible. Even if you find your prior work unremarkable, it may be valuable for the evaluator. 

Back to top

 

Selection Process

Resume screening → Assignment selection (coding test & thematic assignment) → Online interview → Final result announcement

• 4/30 (Fri): Announcement of document selection results and Assignments guidance
• 5/9 (Sun), 23:59 PM: Deadline for submission of assignments
• 5/24 (Mon): Announcement of the results of the assignment selection and guidance for the online interview
• 7/9 (Fri): Final result announcement

 

Back to top

 

Salary and Benefits

Salaries

• 2,500 yen an hour for technical college, university, graduate school students
• 2,000 yen an hour for high school students
• Work hours：8 hours in principle. 5 days a week excluding Saturdays, Sundays, and public holidays.  

Others

Earphone allowance: 3,000 JPY

 

Back to top