Live Updates
Latest from LinkedIn
Vatsal Patel
Principle Research Engineer | AI in healthcare
Real-world deployment is where technology and perception truly get tested.
At WHX Dubai, our team at HawkFranklin Research conducted live previews of Derma
AI, a comprehensive dermatological skin lesion detection platform we developed in
collaboration with Oracle Health and EHSUAE.
Seeing the system perform seamlessly under live, unscripted conditions was a surreal and proud
moment. We designed Derma AI to bridge the gap between patient anxiety and clinical efficiency
using multi-instance learning (MIL) models for detailed differential diagnosis.
Vatsal Patel
Principle Research Engineer | AI in healthcare
Stop renting intelligence. Own it locally.
Two years ago, running a 1B+ parameter LLM on a phone was a fantasy. Today, it’s working code.
I’m excited to share AURA, our first open-source app from HawkFranklin
Research.
We are using LiteRT to run heavy models like DeepSeek R1 and Gemma 3 directly on Android
microprocessors. Your query remains private. No data sent to OpenAI/Google/Anthropic. Total
offline capability.
Vatsal Patel
Principle Research Engineer | AI in healthcare
For the past year, I've been deeply involved in building the next generation of cancer
diagnostic tools. Developing OncoGemma, our Advanced Pathology Workstation, has
been uniquely challenging.
OncoGemma integrates high-resolution Whole Slide Imaging (WSI) with multimodal AI... structured
around:
1. Diagnosis: Cancer Subtype Prediction
2. Prognosis: Mutational Biomarker Identification (Paired with my proprietary Graph Attention
Autoencoder for risk stratification)
The Journey
Experience & Education
Principal Research Engineer
HawkFranklin Research
Developing end-to-end machine learning and AI projects for computational biology. Created an Agent-based framework for privacy-preserving model training.
Mar 2023 - Feb 2025
M.Sc. Artificial Intelligence
IU International University of Applied Sciences
Graduated with a GPA of 1.8 (German scale). Focus on advanced ML, deep learning, and biomedical applications.
Machine Learning Analyst
BRIC-National Institute of Biomedical Genomics
My foundational First Job experience where I designed deep learning frameworks. Developed automated DL pipelines on GPUs, Convolutional Autoencoders for Histopathology, and Graph Attention Networks for Biomarker Identification.
B.Sc. Mathematics
St. Xavier's College
Grade: 9.3/10. Major in Mathematics with a Minor in Physics. Established the analytical and rigorous problem-solving foundation necessary for advanced algorithm development.
Specializations
Research & Certifications
Cancer Genomics Works
Deep Learning models applied strictly to oncology. Key focus areas include Immunomethylomic Tuning of Pathology to Text Models for HNSCC Risk Prediction, and utilizing Graph Attention Networks to integrate diverse multiomic data types from TCGA for biomarker extraction.
View Project DetailsMultiOmics & Biomarkers
Visualizing Graph Attention Autoencoder extracting critical representations from high-dimensional oncology data. Integrates genomics, transcriptomics, and epigenomics into a unified latent space.
Explore Sub-Projects
Deep Learning & TF Certification
DeepLearning.AI TensorFlow Developer. Official verification in building, training, and deploying scalable neural networks for complex visions and signal data.
Research Papers
- Top In-Silico pathogenicity prediction methods (BioData Mining, Jan 2025)
- Dysregulated immune response for COVID-19 (Cytokine, Dec 2023)
BioData Mining | Abstract
Company Portfolio & Tools
HawkFranklin & Flagship Projects
As part of HawkFranklin Research, my work extends into comprehensive company ecosystems and open-source intelligence.
OncoGemma
Advanced Pathology Workstation integrating high-resolution WSI with multimodal AI for cancer subtype prediction.
PelliScope
Dermatology AI. Pre-consultation workflows powered by differential diagnostics with an AUC ≈ 0.863. Focus on GCC & US Telemedicine.
AURA
An intelligent patient vitals and timeline tracker. The first open-source application from HawkFranklin Research aimed at modernizing clinical data flows and enhancing patient monitoring interfaces.
MedDiscover RAG
An AI-powered RAG tool for biomedical researchers to query PubMed using MedCPT GPU embeddings and LLMs.
HuggingFace / GithubML Copilot Agent
Interactive machine learning assistant streamlining data preprocessing, training, and evaluation via LLM command-line interface.
LangChain / PyPI