The Berkeley Science Fellows Program: On campus professional development opportunity for postdocs



The mission of the program is two-pronged‐ provide career transition opportunities to postdocs by helping them experience working with startups; 2. Helping startups solve scientific problems and reach their milestones while being incubated on campus.

Our team

Naresh SunkaraEvangelia VamvakaEduardo González GrandíoAriani Wartenberg, and Gaia Andreoletti


The core function of the Berkeley Science Fellows Program is to connect postdoctoral researchers with startups in the UC Berkeley entrepreneurial community. This will provide valuable real world working experience to postdocs while helping the startups benefit from the scientific expertise of the postdocs.

Berkeley Science Fellows will offer their advice to startups to solve specific scientific challenges including product development, process trouble shooting, grant writing/reviews, experimental data analysis, and presentations, to participate in discussions in brainstorming sessions to develop start up ideas.

The Berkeley Science Fellows Program will be responsible for recruiting postdocs (after an interview) and present them to startups at SkyDeck. Science Fellows will be given an opportunity to select the startup they would like to work with as well.

Qualification: Candidates must be a postdoc at UC Berkeley to be eligible to participate.

Time Commitment

It is understood that postdocs have a primary commitment to their research at UC Berkeley.  Science Fellows will need to commit 4-5 hours a week to assist the startup. The hours are flexible and may include evenings and weekends. Please note that the Science Fellows are not compensated for their volunteer assistance.

Because postdocs will gain valuable experiences while working in a fast paced startup environment, the selected postdocs are expected to deliver on their agreed-upon time commitments. To this regard, the Science Fellows will be intermittently contacted by the Program leads to assure that all commitments are being met.  Additionally, there will be monthly participant check-ins to solicit feedback to improve the Berkeley Science Fellows Program. 

Benefits to postdocs

- Gain experience working with startups.

- Get exposure to potential employment opportunities

- Add work experience to your resume

- Receive a letter of recommendation from a startup

- Serve on the Science Fellows Advisory Board (after completion of internship)

- Receive satisfaction of helping a startup succeed

Benefits to startups

- Access to world class talent (free of cost)

- Future Employee pipeline

- Help with troubleshooting specific problems

- Help with grants (review/write)

This is an unpaid (volunteer) program.

Please use this link to register as a Science Fellow:

If you are a startups trying to recruit postdocs as Science Fellows, use the link below to register: 

 Skydeck Startups recruiting Berkeley Science Fellows

Current postdoc consulting opportunities:

Voyage Biomedical

Cooling the brain to preserve it during the medical emergencies

 Problem being addressed 

We need help developing a model of heat flow in the brain. When we send in a certain amount of coolant at a controlled flow rate, we want to be able to predict how much cooling we will achieve. This will allow us to refine the method of treatment used for our medical device. 

Technical skills needed to help solve the problems.

MechE, BioE, Physics would all be relevant. Also would love to talk to biochemists and chemists about using different preservative medium to keep the brain colder for longer.


We're developing a new video compression technology called video vectorization, which can reduce the bitrate of video content by up to 98% (compared to h264) while improving quality. It is especially effective for computer generated video content such as animations or screensharing. We're working with video platforms, compressing their video content to improve their video experience while also providing significant reductions in bandwidth&CDN costs

 Technical skills needed to help with to solve the problems.

1)    Computer vision, especially object detection & object tracking, ideally with Python and OpenCV 2) Video engineering and video compression technologies

2)   Get involved! We're writing a few more patents, and would love for you to join and help us co-author. We're a team full of engineers from MIT & IIT, and we are also super generous with equity for smart people 

Specific problems being solved.

We are currently developing a proof of concept of our technology for 2d animated cartoons (like the Simpsons). This involves taking in an input video Split the video into 'scenes' For each scene, detect if the frames in the scene can be reduced to vector graphics For scenes that are vectorizable: Segment image into separate objects (we need to make this run faster) Separate background objects from foreground objects (needs more work) Track movement and morphing of foreground objects over time (needs more work) Find vector graphics representation for each object (we have this down) Find tweening/animation for each foreground object that is moving and morphing (we especially need help with this) Saving all the info into our custom vector graphics file format (We have a hacky implementation, and need to develop a professional file standard) Our player then takes in the vector graphics video file, and compiles it to html5 and svg for rendering on the client device. 


We make it possible, for the first time in the history of bitcoin, to walk into a store, buy a cup of coffee and pay for it with bitcoin even if the merchant doesn't accept it - securely (without giving up control of your keys) and instantly (over the lightning network).

Technical skills needed to help solve the problems.

Embedded C, Physical Unclonable Functions, ARM Cortex M23 Architecture & Bitcoin Lightning Network 

Specific problems being solved.

1. Separation of signing keys and channel secrets from lightning daemon to manage on hardware/app. (Possibly using narrowband IoT or LoRA to broadcast lightweight channel updates from hardware with a light mobile/cloud client?)

2. Evaluating and implementing Physical Unclonable Functions on M23 (Nuvoton M2353) chips for HD wallet key storage.


Lenders & Insurers still don’t have an effective way to get 360 degrees overview of the SME and understand their needs, preferences, habits, and issues in real time – before they call or come to the Lender
Lenders & Insurers can’t track lots of events that are happening with their SME clients in real-time and which are highly important from sales, risk, compliance, fraud, and churn perspective. Thus, Lenders & Insurers lose money because lots of important data which is only available in business systems SME use is passing by.

Examples: SME got less income compared with the previous period; Paid the taxes; Connected an account of another Lender/Bank; Expenses are growing; Hired new people; Stocks are full; Customer Acquisition costs are growing; The number of website visitors is decreasing; Got the tax fine; New revenue stream through eWallets; 

On top of that, Lenders & Insurers have access to the traditional data sources only which can't reflect whole business status and lifestyle of SME. Lack of overview leads to bad interaction with clients and low level of the support to their needs, issues, preferences, and even habits

Technical skills needed to help solve the problems.

We need the people with a deep understanding of the processes in matching and structuring the data from various sources. Analyzing this data from behavioral, scoring, risk and financial assessment points of view.

 Specific problems being solved.

We need the people that can help us
1) Create the valuable visualization of the data from CRM, ERP, Accountancy and othe systems
2) Build ML, scoring models based on this data
3) Create the unified data matching and mapping chemes
4) Create the ML models so that the decision making system could be self-educated

Pitch video

 Cura Therapeutics

Cura therapeutics is developing innovative immunotherapies to cure a wide range of cancers, specifically solid malignancies. Our technologies harness cytokines and cytokine receptors to create multi-functional proteins with potent anti-cancer properties. 

 Technical skills needed to help solve the problems.

Post-doc with expertise in protein modeling, 3D molecular design and bioinformatics method for predicting three-dimensional structure model (I-TASSER, SWISS-MODEL or others)

Specific problems being solved.

Predict 3D structure and conservation domains of new chimeric proteins

Flux technology

Flux invented a new material for gas separation. Starting points is MOF nanoparticles suspended in polymer solution. To make hallow-fiber membranes we will need to increase viscosity of the solution. However, increasing solution can result in phase separation between nanoparticles and polymer matrix. The problem statement is: How we can increase viscosity of our solution without phase separation? 

Technical skills needed to help solve the problems.

Polymer chemistry, colloidal chemistry.

Specific problems being solved.

Main: Increase viscosity of our solution without phase separation 

Additional: Increase plasticization resistance of our polymer composite 


 We are solving the commute problem in a dense cities by running network of fixed route shuttles. The shuttles will be equipped with self driving car sensors such as Radars, LiDARs & Cameras. We want to leverage these sensors and a prior map to enable immersive experience for the riders along with route(s). Some of the experiences include discovery of local business such as coffee shops, grocery stores, hair dressers. Over time we can learn the rider behavior and offer them experiences and discovery relevant to that particular rider through a opt-in mechanism

Specific problems being solved.

Object Classification & Object Tracking

Object Tracking: Tracking dynamic objects in the outside environment such as cars, pedestrians and bikes. In particular, we want to solve the challenge where there are occlusions in the scene and the object we are tracking loses its persistence and gets a new ID when it re-enters the scene

Object Tagging: Creating a RoI for an object using hand gestures in an indoor environment and training the inference mode using voice commands. The robot can take images of the object from different perspectives to train the inference model. The initial goal will be to train the inference model for object tagging using hand gestures. The eventual goal will be to reduce the number of training dataset needed for additional objects that might be similar using techniques such as MAML (Model Agnostic Meta Learning)


Cell separation for life science research and cancer diagnostics.

Technical skills needed to help solve the problems.

Single cell analysis techniques. Particularly, single cell sequencing or flow cytometry, cellular digital image analysis involving AI.

Specific problems being solved.

1) Understand user experience, particularly sample preparation issues, for single cell sequencing workflows.

2) Learn about application of AI to cellular digital image analysis.

C. Light Technologies

C. Light is working to provide objective and quantifiable feedback regarding a patient's neurological health through eye motion measured on the cellular scale, focusing initially on drug efficacy in multiple sclerosis and expanding onto other neurological indications. 

 Technical skills needed to help solve the problems.

Advanced statistics, machine learning, MatLab programming are all essential to the data analytics we apply to the eye motion information we are collecting. Clinical trial experimental design is an ancillary skill that would also be valuable in helping us make the most powerful conclusions from the data we are collecting today and in the future. 

Specific problems being solved.

C. Light has collected eye motion data from over 200 patients with MS and 100 controls. We have been able to correlate eye motion to the severity score (EDSS) of a patient, predicting EDSS with ~70% accuracy. We need help applying statistical models and machine learning to boost the performance of our predictors either by feature extraction or convolutions AI models. We'd also like to know the degradation of performance as the fidelity of the data is degraded as a way to demonstrate a direct competitive advantage to other eye tracking technologies. 


Edentulism - using 3D tech to create personalized bone implantable medical devices. In this case, we are a drill free personalized dental implant system

Technical skills needed to help solve the problems.

*Bone scientist - skills with bone healing studies and evaluation of implantable device surface texturing
*Electrical engineer - Skills in piezo ultrasonics and medical device design/engineering
*BioMaterials Scientist - Skills in bone grafting materials and formulations, Ca or synthetically based materials.
*3D modeling designer/engineer - Skills in 3D modeling software and product workflow.

Specific problems being solved.

*Bone scientist/researcher - We need to define and analyse our surface texturing techniques for our Ti based implants. Expertise with Ti6Al4V surface texturing, uCT, SEM, histology, mechanical loading, animal models to evaluate real time bone healing. Specific projects will be defined at the time of collaboration.

*Electrical engineer - Expertise in piezo ultrasonic electronics. Our bone implants are delivered via ultrasonic vibration. We have a piezo electronic device currently but it needs re-engineering and re-design in to a medical device. 

*Biomaterials engineer - We are interested in evaluating grafting materials to increase initial bone stability with our mechanical retentive elements. We would also like to evaluate various surface texturing methods to enhance initial bone growth. 

*3D modeling software desiger/engineer - Our proprietary design features need specific 3D modeling needs which we would like to design a workflow functionality in 3D modeling software such as Rhino, AutoCAD, or? etc... We are currently using predicated subtractive machining to make out implants but we are also interested in additive manufacturing. nTopology is a software dedicated to additive design and manufacturing. Designing user workflow interface will be very important to our manufacturing process.

Here at we aim to build the future of work where distributed teams working on remote are united under one culture. We aim to help companies decode their corporate DNA by measuring culture health through tracking patterns in the communication channels. Our next step is to calculate culture ROI (culture health mapped to corporate performance). 

Technical skills needed to help solve the problems.

Data Science, ML, python 

Specific problems being solved

We currently grab data from Slack and Zoom. We would need help to participate in building the solution to our Zoom module. We would like to decode culture from voice, transcript and Zoom meta-data together and map analysis to our current model. Also we would love researchers to play with data we grab from Slack to bring extra insights into how to decode Culture DNA.

Orbis AI

People have the unlimited freedom to express themselves online visually, with the help of face filters or visual effects. The same freedom does not exist for voice, locked by biological constraints. We want to make impactful voice available to everyone, as easily as wearing a new t-shirt.

Technical skills needed to help solve the problems.

ML and Deep Learning
Signal processing
Sound engineering
Backend server architecture design and engineering

 Specific problems being solved

Improve our core ML and deep learning engines for faster speech-to-speech conversion, while preserving fidelity and similarity of the cloned voice.

Implement new DL models from recent academic papers for speaker adaptation TTS with emotion embedding, as well as speech-to-speech voice conversion models.

Improve our backend architecture for faster and cheaper inference using ML and DL models.


We develop an AI hardware accelerator IP based on embedded Flash memory that can be built in a standard CMOS technology. The IP can be included in any AI SoC chips to solve power consumption issue that is critical in battery-powered mobile edge devices. 

Technical skills needed to help solve the problems.

Processor architect or a data scientist 

 Specific problems being solved

We are developing Deep Neural Network (DNN) IP that performs energy efficient MAC operation based on Analog Computing-in-Memory (ACiM) architecture. The DNN IP is a core building block for an intelligent MCU we plan to develop. And the MCU will include RISC-V processor along with our DNN IP and other sensors to carry out inference task in IoT devices. To reduce the power consumption of the hardware, we would like to develop a tiny ML model incorporating efficient quantization and pruning algorithms. For this purpose, we want to recruit researchers who can help us in developing this challenging task.


Stip provides a platform for companies that lets them handle and optimize their customer care. Stip’s platform allows to have in the same place e-mails and all social network (Facebook, Twitter, Instagram) messages from the company’s users; it also automates a lot of tasks (like discriminating between caring / no caring content, or categorize the message, collecting users’ missing data) that cause big waste of money, time and resource, via a three-layer artificial intelligence (with deep learning technology), and let the operators only handle the important issue: solving the users’ problems.

 Technical skills needed to help solve the problems.

We need experts in data mining, data analysis and machine learning/artificial intelligence, in order to properly process our clients’ data, find relevant information in them, and categorize them in order to ship the client users’ messages to the appropriate department/operator. Also, we can use those data to perform predictive analytics. Python programming skills would be preferable since our platform is mostly programmed in it. Some experience/knowledge with natural language processing is a nice to have skill.

Specific problems being solved

1.     Caring / No Caring Engine and Clustering from APIs

2.     Scraping of external comments from companies’ official channels (ex: Trustpilot) or from APIs of social listening platforms.

3.     Use of "Caring no Caring" engine to identify critical contents to be managed.

4.    Clustering analysis to identify recurrent issues ordered by frequency..

Advantages: they are used to do clustering analysis, therefore, they should apply something familiar to them, in a new work environment.

Issues: create an AI that is general purpose, not dependent from companies sectors 

 Facile Therapeutics, Inc.

Clostridium difficile infection (CDI) is a life-threatening infection of the colon. While antibiotics can clear the infection, they also kill healthy gut bacteria, resulting in re-infection rates of 25-60% within two weeks. Facile has demonstrated that Ebselen is a potent anti-toxin and is effective in a mouse model of the disease ( Most significantly, Ebselen has been previously evaluated in hundreds of patients in Phase 2 & 3 clinical trials in the US, the UK and Japan (all for indications unrelated to infectious disease) and shown to have a favorable safety profile. Facile’s goal is to bring this drug directly into a proof-of-concept clinical trial.

Technical skills needed to help solve the problems.

Ability to review and summarize the scientific literature based on the principles of drug action, PK and safety is essential. 

Specific problems being solved

Facile will have a pre-IND meeting with the FDA to propose bringing Ebselen directly into a Phase 1b efficacy trial without generating pre-clinical safety data. A key part of our request package is a detailed review of the scientific literature on the mechanism of action, pharmacokinetics & pharmacodynamics, safety /tox and efficacy of Ebselen. We need technical assistance putting this review together.

XCloud Networks

Making on-prem physical data-center networking as easy as cloud

 Technical skills needed to help solve the problems.

Fundamental low level, network protocol level, math, research and development skills

 Specific problems being solved

Control plane, data plane for packet forwarding over BGP unnumbered adjacencies

 Coreshell Technologies, Inc.

Coreshell is introducing novel materials into Lithium-Ion batteries to improve energy density and lifetime

Technical skills needed to help solve the problems.

 Synthetic chemistry

Materials characterization


Specific problems being solved

We are looking for a Chemistry or Materials Science post-doc with experience in solution-phase synthetic protocols for inorganic or organic materials. Ideal candidates will have experience in both. More specifically, we are interested in synthetic techniques for forming thin films of materials using solution-phase processes (such as chemical bath deposition, electrodeposition, etc.). The goal of the project is to initially demonstrate thickness control, uniformity and conformality of thin films of specific candidate materials on test substrates (such as Si wafers), followed by transfer of these coating processes to lithium-ion battery electrodes. Ideal candidates could have experience in sol-gel chemistry, nanoparticle (or other inorganic) synthesis, polymer synthesis and/or grafting of polymers onto various substrates. In addition, some experience using standard materials characterization techniques (such as AFM, SEM/TEM, XRD, Raman/IR spectroscopy, etc.) is desired. Experience with Lithium-Ion battery materials and electrochemistry is also a plus.

Given the small weekly time commitment (4-5 hours/week), initially the role of the post-doc will be entirely advisory- coordinating with the CTO to define synthetic and characterizing protocols for various materials- with the CTO and Coreshell staff to then conduct the proposed experiments themselves. Later, depending on interest and availability of the post-doc, there will be potential for the post-doc to perform certain characterization tasks by himself/herself at the Molecular Foundry at LBNL through the company's user proposal. In addition, the Coreshell team and post-doc will work together to evaluate the performance of coating materials when applied to lithium-ion electrodes through a variety of electrochemical tests.

Automating repetitive tasks in a small business environment

 Technical skills needed to help solve the problems

Machine Learning, AI

Specific problems being solved

We are exploring to transfer Q learning concepts and methods to task completion in a business environment. 

Nexilico, Inc.


Problem Being addressed

Nexilico, Inc. is a computational biology startup offering a first-of-its-kind predictive solution at the intersection of precision medicine and microbiome research with the ultimate goal of predicting individual-specific response of the human gut microbiome to therapeutics and leverage that information to improve drug development. Our computational platform provides a virtual, cost-effective approach to (1) predict subject-specific efficacy and toxicity for pharmaceuticals and (2) personalize microbiome-based therapeutics (also known as bugs as drugs).


Technical skills needed to help solve the problems.

Bioinformatics, Computational/Systems Biology, Programming (C++/Python), Microbiology


Specific problems being solved.

Computational analysis of metagenomic (and metaproteomic) data in the context of human gut microbiome