Shalini Ananda
San Francisco Bay Area
Summary
Experienced machine learning researcher and instructor with 10+ years of applying advanced deep learning techniques to healthcare, drug discovery, NLP, Generative music, and medical imaging problems.
Overview
19
19
years of professional experience
Work History
ML Reseacher
Scale AI
02.2023 - Current
- Frame tasks and guide large language models, iteratively refining prompts to optimize model performance
- Modular prompt architectures break down complex problems into simpler sub-prompts
- Tasks were relevant to code generation in C++ and Python.
AI course instructor
UCSC Silicon Valley Extension
02.2022 - Current
- This course explores using LLMs for multimodal sustainability applications: - Forecasting renewable energy production by fine-tuning transformer models on time series data
- Students will learn full-stack development for integrating LLMs into end-to-end systems
- Assignments will involve training models with frameworks like HuggingFace Transformers and deploying them via REST APIs and web interfaces
- Datasets will come from sources like NASA and sustainability-focused Kaggle competitions.
Engineering course Instructor
UC San Diego Extension
11.2021 - Current
- Instructing computer science courses online - - Python and C++ - Created a new class called "Growing food on Mars" (systems development)
Founder
Anandamide Sound Machine
06.2019 - Current
- I developed prototypes for algorithmic music creation using techniques like GANs, VAEs, and transformer networks in PyTorch
- I leveraged GCP services like Cloud TPUs, Cloud Storage, and BigQuery to train models on large music datasets
- The prototypes allowed rapid iteration to evaluate different model architectures and techniques
- Core technologies used include PyTorch for model development, React for web app prototyping, and GCP for cloud-based model training and deployment
Machine Learning Specialist
EdCast
02.2022 - 02.2023
- Building models for upskilling
- Create a model uses tree augmented naïve bayes classifier to predict the future performance of an employee and this makes the predictions easier to explain.
Visiting Researcher
UC San Diego
06.2020 - 10.2021
- Predicting and treating relapse on recovering opioid abusers using machine learning
- Worked with the Department of Neuroscience/ Psychiatry.
NLP engineer
Sensentia
10.2018 - 10.2019
- In 2019-2020, I led development of an ML system to automate medical billing code assignment
- We leveraged TensorFlow and PyTorch to train graph neural networks on insurance claims data and CPT code networks
- The GNN models inferred probabilistic linkages between patient symptoms, procedures, and billing codes
- We continually improved models by connecting to partner EHR systems via FHIR APIs and retraining on new patient data
- Our work reduced manual errors in medical billing by 90%, helping clinics improve care quality through automation.
Co-Founder & Chief Technology Officer
CRIXlabs, Inc. (DBA Quantified Skin)
01.2013 - 09.2019
- As the CTO, I managed model development and deployment for mobile skin diagnosis with large pharmaceutical clients
- We leveraged OpenCV and early TensorFlow to build and train convolutional neural networks on a dataset of dermatological images
- The models were optimized for fast inference on mobile devices - key in those early days of mobile AI
- For algorithm interpretability, I implemented techniques like saliency maps and LIME to provide visual explanations alongside model predictions
- These model introspection techniques were critical for building trust and adoption in nascent CV models
- I authored patents covering our unique multi-ethnic skin condition analysis and personalized product matching system
- This work required curating and carefully protecting a diverse skincare image dataset
- On the engineering side, we containerized models with Docker for reproducible deployments from prototype to production
- Kubernetes was just taking off, so we relied more on Chef and Puppet for cloud orchestration
- For simulation, I leveraged C++ and early CUDA libraries to implement pharmacokinetic models of drug reactions
- We were limited to CPU and early GPU clusters for distribution modeling.
Rock Health Fellow
Rock Health
06.2013 - 12.2013
- Stanford University
Postdoctoral Researcher
UC San Diego
10.2012 - 12.2012
- Co-inventor of NanoSim : A toolbox for a NMR based CMOS biosensor Started in Sam Gambhir's lab (Multimodal Molecular Imaging Lab)
Graduate Researcher
UC San Diego
09.2008 - 10.2012
- T2 Tunable Porous Silicon Iron oxide nanocomposites for Magnetic Resonance Imaging guided drug delivery : Monte Carlo modeling to determine contrast generated from new materials on MRI
Staff Research Associate at Magnetic Resonance Lab, Hillcrest
UC San Diego
01.2005 - 01.2007
- Looking at Gadolinium and iron oxide base contrast agents to enhance visualization of scars in liver fibrogenesis.
Education
PhD - Materials Engineering
University of California, San Diego
San Diego, California01.2012
Masters of Science - Applied Mathematics
University of Sydney
Sydney, Australia01.2005
Bachelor of Science - Applied Mathematics
University of Sydney
Sydney, Australia01.2004
Skills
- Machine Learning
- Model Development
- Machine Learning Integration
Honors Awards
San Diego Venture Group COOL COMPANY 2013
Patents
- Closed Loop Feedback System for Processing Data Values Representing User Behavior to Provide Product Recommendation Using Deep Belief Networks
- Method and System for Predicting Spatial and Temporal Distribution of Drug Carriers
- Method and system for using 1D, 2D, 3D interaction profiles between small molecules and cells to predict spatial and temporal distribution of molecules
- Method and system to capture dynamic user information via healthcare apps collected over a continuous interval
- TagGen: Automatic tagging of data generated from text, images, video and sensors for personalized product recommendation using deep neural networks
Timeline
ML Reseacher
Scale AI
02.2023 - Current
AI course instructor
UCSC Silicon Valley Extension
02.2022 - Current
Machine Learning Specialist
EdCast
02.2022 - 02.2023
Engineering course Instructor
UC San Diego Extension
11.2021 - Current
Visiting Researcher
UC San Diego
06.2020 - 10.2021
Founder
Anandamide Sound Machine
06.2019 - Current
NLP engineer
Sensentia
10.2018 - 10.2019
Rock Health Fellow
Rock Health
06.2013 - 12.2013
Co-Founder & Chief Technology Officer
CRIXlabs, Inc. (DBA Quantified Skin)
01.2013 - 09.2019
Postdoctoral Researcher
UC San Diego
10.2012 - 12.2012
Graduate Researcher
UC San Diego
09.2008 - 10.2012
Staff Research Associate at Magnetic Resonance Lab, Hillcrest
UC San Diego
01.2005 - 01.2007
PhD - Materials Engineering
University of California, San Diego
Masters of Science - Applied Mathematics
University of Sydney
Bachelor of Science - Applied Mathematics
University of Sydney
Similar Profiles
Paul KellerPaul Keller
GEOSPATIAL AI TRAINER II at SCALE AIGEOSPATIAL AI TRAINER II at SCALE AI
Marcel KingMarcel King
QA Team Lead at Scale AIQA Team Lead at Scale AI
Abdelrahman FaroukAbdelrahman Farouk
Queue Manager at Scale AIQueue Manager at Scale AI
Marwa AboubakerMarwa Aboubaker
Machine Learning Data Specialist at Scale AIMachine Learning Data Specialist at Scale AI
Bikashita (Leema) SaikiaBikashita (Leema) Saikia
Senior Manager, Legal at NielsenIQSenior Manager, Legal at NielsenIQ