Gain access to top talents from reputable universities globally to work on your Deep Tech projects with co-funding of up to SGD$3,000/month from SGInnovate. These highly-skilled talents will bring in fresh perspectives that add value to your organisation. Increase your brand awareness by being a part of the programme’s outreach activities.
Head of Delivery (Singapore), Taiger
Data Platform Lead, ViSenze
Access to Top Talent
Attract top Deep Tech talents – undergrads and PhDs from reputable universities across the globe.
Co-Funding
Attract top talent with an award of up to SGD$3,000/month from SGInnovate, in addition to the stipend provided by your organisation.
Brand Outreach
Increase your brand awareness by being a part of the programme’s outreach activities.
Fresh
Perspectives
Bring in energetic students with fresh perspectives to test out new ideas or roles for your organisation.
Networking
Opportunities
Get plugged into the Deep Tech ecosystem through exclusive invites to various industry and community events.
Company Profile
Your organisation should be a startup specialising in Deep Tech with proof of financial stability.
Deep Tech Projects
Projects offered must be related to Deep Tech, such as Artificial Intelligence, Blockchain, Cybersecurity, Quantum, and Robotics
Skilled mentor
Mentors should demonstrate technical expertise, ideally with five years of technical and two years of mentorship experience.
Please see the FAQs for more details and eligibility criteria.
Click here for FAQs
TAI1: Speech to Text Processing Using NLP
This project converts speech to text for Singlish and English.The apprentice is required to experiment and develop a module or a pipeline of modules that will take speech as an input and return the transcript for the speech.
TAI2: Contextual Data Extraction Using CV and NLP
Identifying and extracting the inherent context of the document is a significant step in automatic information extractions tasks. The objective of this project is to develop different modules,such as de-noising, de-skewing, and to enhance the image quality of a scanned image. This scanned image will be processed using OCR to extract relevant information using NLP.
POR1: Dynamic Cargo Pricing Through Machine Learning
90% of the world trade travels via maritime, at least once in its lifetime. Portcast wants to apply Machine Learning to the extensive datasets from the maritime industry that it captures and convert it into actionable alpha intelligence. The project scope is to transform the raw datasets (from economic indices, geospatial data, to news/sentiment information) into insights and patterns that would be interesting not just for the logistics industry, but beyond that to finance and insurance companies.
KON1: Predictive Analytics Using Machine Learning for Businesses
The purpose of the project is to build a fast and reliable time-series forecasting pipeline for sales data. This project will give the apprentice an opportunity to work on an essential financial dataset, and evaluate internal and external factors affecting sales. This project will require the apprentice to assess various forecasting engines to identify the right model based on the data for our customers. The main aim of the product will be to provide our customers the most accurate forecast with actionable insights.
KON2: Information Extraction Using Natural Language Processing
We are developing an automated system which requires zero configuration or any upfront annotation using the latest developments in the fields of CV and NLP for information extraction. We are building a next generation human interaction system which will be as natural as asking questions to an in-house analyst who knows all the answers. We are doing this by using state-of-the-art DeepLearning research in NLP to understand users’ intention and entity from the natural language to give the most appropriate answer.
ITX1: Information Extraction In the Legal Domain
At INTELLLEX, we handle millions of legal documents, and we want to build an automated end-to-end system to extract useful information from the legal documents.For each document, we need to identify the areas of law being discussed and legal keywords being used.
The apprentice will be involved in the development of novel ways to tackle search, data extraction, and the building of a large and ever-changing knowledge graph. He/she will participate in the creation of sophisticated pipelines handling everything from data exploration, pre-processing, running advanced Machine Learning models, and incorporating user feedback in the training loop while running in production.
AFF1: Social Media Inference Through Deep Learning
At Affable, we profile millions of users every day and use Machine Learning to infer things such as their interests, age, and location. We use a host of signals to make such inferences, including but not limited to their profile pictures, posts, stories, and captions.
This project involves conceptualising, training and deploying Machine Learning models that can infer even more characteristics about social media users. Some of the traits we want to extract further are psychographics, education level, job function, and income level.
The main challenges of this project are two-fold: volume and speed. We are tracking over 1.5 billion social media posts and 300 million social media users. Machine Learning at this scale is challenging, time-consuming but also very rewarding. The apprentice will get first-hand experience in building models that are not only accurate but also fast and production-grade (some of our models run inference on over 20 million images per day).
DAT1: Machine Learning for Personal Identification Data Linking
In May 2018, the new General Data Protection Regulation (GDPR) came into effect. It is essential for all businesses, and the Dathena Compliance is created to help companies to be compliant. The highest priority for the data protection solution is to provide sensitive data extraction, detection, and structurisation. One of the challenges is to link person names and other Personal Identification Information (PII) like their address, email, and card number.
We aim to leverage state-of-the-art methods to find the most weighted pairs of PIIs linked to the person. Unstructured text data is used in this task for information retrieval and matching. The main idea of the approach is to evaluate three degree-based measures to find the most suitable PII node for the target name mentions. Experimental results on domain-dependent datasets show that the graph-based method performs comparably with state-of-the-art techniques.
DAT2: Unsupervised Documents Classification for Asian Languages
One of the main tasks of the Dathena99 product is the classification of the huge amounts of unstructured financial data into business categories and different levels of confidentiality.
The main challenge faced is the lack of labelled data. Because of this, the first step of the Machine Learning pipeline is applying unsupervised techniques for the documents. This step could be replaced by automatic cluster labelling by using keywords or relevant concepts to the group of the documents.The most suitable label should not only explain the central theme of a particular cluster but also provide a means to differentiate it from other clusters in an efficient way.
The main challenge of this project is scaling the English based approach in other Asian languages such as Chinese, Korean, Thai, Japanese. We are proposing to use Apache Spark for distributing computations, Scala as the primary programming language, and HBase as a storage database.
QUO1: How to Help a Machine to Ask Good Technical Questions with NLP
We build a composition of models that extract relevant context and intelligence from code and metadata, and then present it to the user in different forms of technical documentation. Technical documentation in the form of questions and answers on platforms like Stack Overflow has become an essential body of knowledge for sharing the intelligence and solving developers' technical problems.
However, it takes a significant amount of effort for a developer to query, digest and review relevant information returned from search engines.Besides, questions and answers could lose their relevance with time, as new versions of technologies make such information obsolete.
Our solution captures the functionality of code while it is being produced and pushed into production. In this project, the apprentice will helpwith translating source codes to natural language in the best possible way.
The apprentice will own a Deep Learning project from beginning to end. He/she will be responsible for defining an implementation strategy after the literature review, implementing the strategy in consultation with medical consultants and assisting in the deployment and validation of the solution.
MUS1: Improve Learning Speed of Music Tagging and Categorisation
At Musiio, we recognise that an accurately tagged database has a multitude of benefits. Besides improving the catalogue’s recall accuracy and user experience as a whole, it supercharges search, navigation and curation for any user. Manually tagging music is repetitive, time-consuming and not without safeguards against listener bias or fatigue. This project focuses on the optimisation of data used for model training, as well as the hyperparameters for our model-training workflow. Achieving excellent performance is critical to deliver cost-effective products.
MUS2: Source Separation for Music Tracks using Machine Learning
Source separation is the ability to take a combined audio waveform and splitting it up into its components. Being able to do this while maintaining excellent audio quality is very difficult.
INT1: Automation of Video Interview using NLP and CV
At Interviewer.AI, we focus on building the penultimate step to hire great talent. Project Phantom is designed to identify key soft skills required for a job based on scientifically-provenand structured interview techniques. This project will use video interview data with labelled models to identify critical soft skills needed for essential customer-facing jobs in Singapore. These are usually non-technical jobs that require traits like interpersonal communication/ learning ability/adaptability and are hard to assess from a CV/ Resume. Once we can evaluate these skills, we can support a wide range of roles for hiring.
SOY1: AI Recommenderfor Fashion Content
Our Machine Learning tech stack primarily uses Python. We use Pytorch and Tensorflow for image processing and recommender systems. We use techniques such as sequence modelling for click-through prediction, and transformer models for text processing. We also use the classic python stack of SciKit Learn, Pandas, and Numpy. AirFlow is used as our distributed scheduler, as well as AWS Kinesis and Lambda for real time processing.
TEL1: Video Analytics for Sports Broadcast using ML and CV
The ability to quickly repeat and slow down sport’s finest and most controversial moments is a paramount feature in every sports broadcast. Although multiple technologies offer the hardware and the systems in instant replay for clipping, cutting, and editing, there is still a need for many experienced operators to do those tasks in a very fast pace manually. Thousands of clips of different lengths are created during a live match. Untagged or minimally tagged piles of clips can quickly become a logistical nightmare when it comes to retrieving the right content.
An ensemble of Deep Learning architectures and advanced Computer Vision techniques are used in this project to accurately recognise players, objects, actions, and events in every clip created during a sports broadcast. The resulting smart engine enables more efficient discovery of outstanding or even non-compliant player actions beyond human-level visual abilities.
BLO1: Clustering & Standardisation of Anime Merchandise Data Using Machine Learning
BlockPunk is building a marketplace for online anime merchandise with authenticity tracked on the blockchain. We are looking for high potential engineering students with background in Machine Learning. As a part of the project, the apprentice will be responsible for:
BYT1: Video Understanding with Visual and Linguistic Cues
In this project, we aim to derive generic visual-linguistic representations for video understanding tasks. It has been shown that linguistic modality provides important complementary information to video understanding. The previous standard approach of combining Computer Vision and NLP is to pre-train networks for visual and linguistic modalities respectively and finetune them for the visual-linguistic tasks. We seek to learn the connections between the two modalities more effectively.
DAS1: Apply AI to Estimate Site Revenue Attribution (SRA)
One feature of our network planning products, “Site Revenue Attribution (SRA)”, helps in determining customer ARPU (average revenue per user). We can estimate the revenue that each cell and site contribute based on each subscriber's ARPU and usage, which then optimises capex allocation using subscriber-specific data and mobility intelligence.
The current solution is implemented using Spark-Scala but is limited in scalability due to the large data size. One goal for this project is to leverage hybrid MPI & GPUs to improve scalability. The other goal is to investigate using a hybrid Graph Deep Learning (GDL) approach.
DAS2: Population Segmentation and Intent Mapping
Digital Out-of-Home (OOH) advertising is becoming more popular. However, the current challenge is that available data is static and unable to show finer Spatio-temporal details. We are working on algorithms to crunch mobile network data into dynamic, in-depth insights, based on actual movements of millions of users, using information such as demographic information, time of day and trip frequency.
This project blends various datasets to define granular segmentation of a population beyond high-level demographics. The apprentice will then build on top of the stack functions to predict intent segments of mobility patterns across geo-locations at various times of the day. This will help media advertisers to plan and develop insight-driven campaigns with access to dynamic human mobility data. Another goal is to leverage a hybrid MPI & GPU approach to handle the large dataset.
Apply AI to Estimate Site Revenue Attribution (SRA)
One feature of our network planning products, “Site Revenue Attribution (SRA)”, helps in determining customer ARPU (average revenue per user). We can estimate the revenue that each cell and site contribute based on each subscriber's ARPU and usage, which then optimises capex allocation using subscriber-specific data and mobility intelligence.
The current solution is implemented using Spark-Scala but is limited in scalability due to the large data size. One goal for this project is to leverage hybrid MPI & GPUs to improve scalability. The other goal is to investigate using a hybrid Graph Deep Learning (GDL) approach.
Population Segmentation and Intent Mapping
Digital Out-of-Home (OOH) advertising is becoming more popular. However, the current challenge is that available data is static and unable to show finer Spatio-temporal details. We are working on algorithms to crunch mobile network data into dynamic, in-depth insights, based on actual movements of millions of users, using information such as demographic information, time of day and trip frequency.
This project blends various datasets to define granular segmentation of a population beyond high-level demographics. The apprentice will then build on top of the stack functions to predict intent segments of mobility patterns across geo-locations at various times of the day. This will help media advertisers to plan and develop insight-driven campaigns with access to dynamic human mobility data. Another goal is to leverage a hybrid MPI & GPU approach to handle the large dataset.
Video Understanding with Visual and Linguistic Cues
In this project, we aim to derive generic visual-linguistic representations for video understanding tasks. It has been shown that linguistic modality provides important complementary information to video understanding. The previous standard approach of combining Computer Vision and NLP is to pre-train networks for visual and linguistic modalities respectively and finetune them for the visual-linguistic tasks. We seek to learn the connections between the two modalities more effectively.
MAY1: Information Summarisation using Artificial Intelligence
Fund Managers today are overloaded with information. They have to process hundreds of emails, PowerPoints and PDF documents daily, searching for insights that can help them make better investment decisions. In this project, the apprentice will develop an automated system to summarise information and deliver insights to the Fund Managers, using Machine Learning algorithms.
LAU1: X-ray Object Detection Using Computer Vision
In this project, the apprentice will build a reliable X-ray object detection system. The system will be used to identify dangerous goods such as weapons and drugs at airports and customs checkpoints. The project is based on recent advances in Deep Learning and Computer Vision technologies. In this project, he/she should take the initiative to see the product through the entire life-cycle, from proof-of-concept, to design and final production.
HYD1: Clearing and Settlement of Security or Asset-Backed Tokens
The project will focus on the building of blockchain infrastructure and software to facilitate post-trade processes, such as the settlement of security or asset-backed token transactions.
HAS1: Risk Compliance (AML) Traceability Tool for Digital Securities
We are building a smart tool with Machine Learning (ML) models to trace discrete transactions of digital securities and funds across different blockchain and payment systems. This tool will empower financial institutions on the STACS blockchain to easily visualise individual transactions moving through various blockchain networks (i.e. STACS blockchain) and payment systems (i.e. IIN Network) in real-time via a UI dashboard. Automated triggers built into the smart tool will alert business users to suspicious transactions based on a fixed set of parameters fine-tuned by the ML model.
KYB1: Create a DAO (Decentralised Autonomous Organisation) for Blockchain Protocol
One of the main imperatives of Kyber is creating a decentralised governance structure that can facilitate the goal of Kyber becoming a key infrastructure of the decentralised economy and financial landscape. In light of this, Kyber is researching and developing a suitable DAO (decentralised autonomous organisation) model for the protocol. The Kyber DAO will help facilitate protocol governance decisions, such as deciding on proposals and protocol parameters.
Cybersecurity in IoT Edge Security
You will get to work on the state-of-the-art APIs for IoT dashboards; perform analytics and provide Cybersecurity to IoT networks; develop security protocols (SSL) on lightweight network routers/devices; design and develop different security protocols; along with cryptanalysis of various architecture and security protocols for embedded systems. You will also get to work on a broad domain of technologies, providing you with insights on various industry trends.
RES1: Demand Side Management Solution to Improve Energy Efficiency
Resync's demand-side management solution would integrate and improve performance of various loads like HVAC (heating, ventilation, and an air-conditioning unit), EVs, other flexible and controllable loads in the built environment, based on various signatures such as localised energy generation, variable tariff, customer preferences and optimisation. The apprentice would be working with the hardware and firmware team to support them in implementing various communication protocol and designing the hardware. It would involve a multi-disciplinary exposure to C++, embedded systems, and their design as foundation.
RES2: Forecast and Optimisation of Distributed Energy Resources
Optimisation is a critical aspect of our both energy cloud solution. With multiple energy resources such as solar, energy storage, electric vehicles, HVAC, other controllable loads. The operation of these energy assets is based on accurate forecasting of generation and consumption patterns and heuristic optimisation of the energy assets operation. These optimisation results are sent to the real-time controller to take necessary actions and provided as insights on the portal for the users.
TRA1: Radio Frequency Development for Wireless Power Transfer
We are seeking self-motivated Radio Frequency Engineers, who have a strong technical background on system-level electromagnetic, signal integrity, power integrity design, analysis and antenna design.The candidate will be contributing to a new RF-based far-field Wireless Power Transfer (WPT) platform, through patented signal optimisation and beamforming algorithms that will extend the range of WPT to 100 meters and beyond.
FIR1: AI-Based Analytics for the Solar and Semicon Industries
Our first market is the solar industry. We build AI software solutions that are customised for that industry. This involves data integration, smart data fusion and AI data analytics applied on hundreds of thousands of gigabytes of data.This project is of a high priority for our organisation, as it is the main source of revenue. The scope includes technical (software development + data science) as well as non-technical tasks (communicate and serve our clients, including managerial and C-level positions).
ACK1: Structural Health Analytics
Structural Health Monitoring (SHM) refers to the process of systematically analysing the data from existing infrastructures like skyscrapers and bridges and extracting useful information about how they age. Such information enables predictive maintenance of critical infrastructures. We are seeking an enthusiastic individual to contribute to our SHM development efforts. The ideal candidate will be eager to learn-on-the-job. He/she should take the initiative to see the product through the entire lifecycle’s proof-of-concent to design to production.
WAV1: Machine Learning for Material Classification and Defect Tagging
The project is about enhancing an end-to-end AI-enabled asset inspection solution, addressing the specific needs of the built environment sector, which includes automated material classification and defect tagging.The apprentice will get an opportunity to work closely with our experienced team from diverse expertise comprising of Electrical Engineers, Software Developers and Cloud Engineers and will work towards developing deep learning models for well-defined classification and defect-recognition problems and productise trained models.
WAV2: Radar Signal Processing for Millimeter-Wave Imaging
Theapprenticewill work with WaveScan's scientists to develop advanced radar signal processing algorithms for our planned beamforming MIMO radar systems. The apprentice will learn the entire R&D process of building radar-based sensor systems for specific applications. This apprenticeship will help understand the bench to the market development process necessary for corporate R&D jobs. At the end of the apprenticeship, he/she will learn a lot in development areas such as Microwave & Radio Frequency (RF) and Signal Processing.
ENT1: Near-Term Quantum Machine Learning
In this project, you will contribute to the technical development of algorithms and applications for Entropica's quantum machine learning suite. The goal is to design and implement variational quantum algorithms for supervised and unsupervised machine learning, and benchmark their performance against existing classical algorithms. The work will involve the use of both real quantum hardware, as well as classical HPC for simulating and prototyping.
ENT2: Scientific Computation for Quantum Machine Learning Algorithms
In this project, you will help to integrate high-performance computing with Entropica's quantum machine learning suite. The tasks include the development of simulated quantum computing environments using parallelisation and distribution techniques on the cloud.
ATO1: Data Modelling for Gravity-Based Resource Exploration
At Atomionics, we are building sensing technology which performs 1000x better than the current state-of-the-art technology using Atom Interferometry, harnessing the potential of wave-nature of atoms. The current ongoing project is building a portable single-axis atom interferometer which can measure gravity very precisely and help in accurately pinpointing hydrocarbon and mineral reserves helping ease the energy needs of the world. The apprentice will build data models to interpret this new gravity data and derive insights.
ATO2: Atom Interferometry for Gravity Measurements
The current ongoing project is building a portable single-axis atom interferometer which can measure gravity very precisely and help in accurately pinpointing hydrocarbon and mineral reserves helping ease the energy needs of the world. The apprentice will be working closely with the product development team, an interdisciplinary mix with prior experience in Atomic physics, biomimetic underwater robots, high altitude pseudo satellites and balloon satellites, and advised by leaders in cold atom physics and navigation systems.
SPE1: Quantum Communications Modelling
One of the immediate applications of quantum networks is the secure distribution of quantum bits to generate secret encryption keys at distant locations. The underlying quantum processes, namely the generation and detection of single photons and entangled photon-pairs are in principle well-understood. Sophisticated algorithms exist (and are routinely used at SpeQtral) that analyse the quantum correlations and distil a secret key. However, in space-to-ground quantum networks, rapid clock drifts and extremely lossy transmission channels add challenges to the well-established protocols.
SPE2: Quantum Entanglement Source R&D
Quantum networks rely on transmission of quantum information similar to how the internet is built on the transfer of classical data. In quantum networks this is realised through shared entanglement between two locations. Entanglement must first be generated and then distributed, for example through transmission of entangled light particles. A significant challenge in creating a global quantum network is overcoming losses in optical transmission media. An elegant solution is using satellites to transmit photons through vacuum (low loss) and make use of direct line of sight from orbital altitudes to distribute entanglement over thousands of kilometres. SpeQtral’s technology heritage includes deployment of the world’s first entangled photon source in space outside of China, and we are currently developing a follow-on mission to be launched in the next two years.
GAR1: Building Facade Inspection with CV
Building and Construction Authority (BCA) has awarded a grant to a consortium that includes Garuda Robotics to inspect buildings in Singapore using drones. This 18-month project is coming online in tandem with new regulations for all buildings to increase the frequency of facade inspection starting 2020.
MOV1: Computer Vision for Robotics Localisation and Navigation
As a Robotics Engineer in Movel AI, you work with the world's best robotics scientists and engineers to implement mobile robotics solutions for some of Singapore's key industry sectors. You get to experience the early life of a Deep Tech startup and sharpen your software engineering skills by working on real-life and critical robotics systems.You will get a chance to work on the full stack robotics software system and hands-on experience with some of the world's most advanced robots. The challenges you can tackle includes:
IIM1: Outdoor Cleaning Robot Development
This project is to develop an outdoor cleaning robot which can autonomously carry out the cleaning work at an outdoor environment, such as Business Park and residential area. The robot should have the capabilities of environment modelling and understanding, localisation and navigation, as well as collision avoidance and other safety decision making. Currently, we have developed the cleaning robot hardware including mobile platform and various sensor installation, such as 3D lidar, RTK-GPS, sonar and 2D/3D cameras. We have also developed some basic function algorithms such as 3D lidar-based SLAM, global and local path planning. However, to enhance system stability and robustness, we still need to develop more advanced and intelligent algorithms.
We need talented students from the following research areas:
IIM2: Multiple-Pedestrian-Tracking System through CV and ML
An intelligent surveillance system is supposed to identify individuals or objects using reconstructed trajectories and re-identification algorithms, which can be created by observingtrajectories of moving targets. Since these applications are closely related to public safety and human-robot interaction (HMI), it is essential to devise a robust tracking algorithm.
The main objective of this project is to track multiple pedestrians in real-time.The purpose of multi-target tracking is to provide accurate trajectories of moving targets from given observations. The produced trajectories can be used for position prediction or re-identification; then provided into the strategy lever of a robot. This can help to smooth interaction with each other in the crowded environment between human and robot. For instance in autonomous vehicle application, it prevents traffic accidents by predicting movement of pedestrians or vehicles.
POL1: Flight Planning for Pollination Drones using ML
In this project, the apprentice will be involved in developing our first autonomous drone platform for pollination-as-a-service, which will be deployed in our paid pilots. To execute this, you will be working on a nano-drone platform on the following areas: control systems, perception, state estimation and motion planning.
Automation of flight for a nano-drone is a non-trivial problem as there are challenges in onboard computational and payload capacity. In this project, the apprentice will get the chance to dive into various stages of the value-chain of automation starting from scalable solution for state estimation to motion planning around complex structures such as plants.
SEP1: Polymeric Membranes Fabrication
SEPPURE creates sustainable nanofiltration solutions to separate chemical mixtures at a molecular level with minimal energy use.Up to 15% of the world’s total energyis spent on industrial-scale chemical separations. This is economically and environmentally inefficient. SEPPURE is disrupting the energy-intensive chemical separation processes, such as distillation and evaporation, through the introduction of novel chemical-resistant nanofiltration membranes. Our sustainable process does not use heat, thus lowering global energy use, emissions, and pollution substantially.We are currently working on several projects to fabricate robust nanofiltration membranes for solvent separation and design membrane systems for vegetable oil, pharmaceutical, petrochemical, and fine-chemical industries.
EKO1: Deep Learning on Echocardiograms
We are building out a global data network housed in some of the world's preeminent imaging labs and academic institutions. This data serves as the foundation for Machine Learning and statistical analysis for disease prediction studies. We are collaborating with some of the world’s top cardiovascular researchers to structure, analyse and publish this data while incorporating the algorithms into a SaaS-based product for clinical use.
You will be required to support and participate in Machine Learning trials, and to organise and analyse large volumes of echocardiograms. There will be opportunities to gain experience with a variety of Deep Learning classifications and segmentation techniques in collaboration with some of the world's premier cardiology centres.
QRI1: Analysis of Human Biopsy Samples with Computer Vision
Cancer is one of the leading causes of death in the world and affects patients and their families for extended periods. Although there has been a significant amount of progress in treatment options, early diagnosis is still the main factor in determining a patient's prognosis. Qritive's mission is to enable doctors to diagnose cancers faster and more accurately, and consequently improve the patient's chances in fighting these diseases.
To achieve this goal, Qritive is leading several joint projects to build these systems. The project collaborators are the two most prominent hospitals in Singapore with a panel of expert clinical researchers and clinicians involved in these projects. Qritiveis in charge of the research and development (both technical and clinical), data collection, and annotation.Clinical validation of the products will be done by the hospitals.
The project is based on recent advances in Deep Learning and Computer Vision technologies. The goal is to reduce the time spent by doctors on the analysis of cancer tissue images and reduce the number of misdiagnoses.
LUC1: Building an AI Model for Interpretation of Cancer Variants
This project involves building anArtificial Intelligence model to power the interpretation of cancer variants in a comprehensive way. At Lucence, we have aggregated all public cancer-variant related databases, and we have also curated our knowledge base through our service to the patients. All this information can be better integrated to provide an in-depth and comprehensive interpretation of genetic mutations in cancer patients.
Through this project, the apprentice will get access to real genomic big data, and participate in the data cleaning and building of databases. These efforts are essential in contributing to better treatment selection for cancer patients. As recent mutation-related treatment information is mostly published in literature, the application of Natural Language Processing skills would be required.
MED1: Deep Learning on Retinal Images
We have gathered a database of about half a million retinal images and are continuously working with leading ophthalmologists.In this project, the apprentice will work on a screening solution starting from literature review to customer deployment. He/she will design a Deep Learning-based algorithm to detect a specific eye condition, working closely with the VP of AI as well as medical consultants.
KRO1: AI for Chronic Wounds Classification
The first step in using KroniKare’s solution is to assess a patient’s chronic wound using the scanning device. This wound is then crossanalysed against more than 15 years of data that has been integrated within the Artificial Intelligence(AI) server. Within 30 seconds, a report is generated with a detailed assessment of the wound, allowing healthcare practitioners and patients to make quicker decisions about the best care and treatment needed.
Please fill in your details below to find out more about the organisations and their projects. You would need to disable your pop-up blocker, if any, in order to download the PDF.
Open for Project
Submission
Publication of
Selected Projects
Open for Apprentices
Application
Shortlisting of
apprentices
Start of the Summation
Programme
There are two runs of the Summation Programme available every year. If you are unable to participate in the current run, our next intake would be in Dec 2020. Applications open May 2020.
Two apprentices with SGInnovate’s Summation Programme discuss their apprenticeships at Singapore-based startups Taiger and See-Mode Technologies.
Shen Yue, a recent apprentice with SGInnovate’s Summation Programme, shares her experiences as an apprentice at Singapore-based artificial intelligence company Dathena.
SGInnovate’s Summation Programme alumnus Gabriel Wong shares how the apprenticeship has helped him better understand the field of deep tech.
Topics: A.I., Data Science / Data Analytics, Startups, Talent