For employers

Hire this AI Trainer

Rafid

Agency

Founder | CEO | Principal ML/Ai Architect | Immunome AI Biotechnologies

New York, Usa

Immuneome-ai.com

Key Skills

Software

Internal/Proprietary Tooling

Other

Top Subject Matter

Computational Biology & Bioinformatics: Deep expertise in engineering

analyzing multi-omic data pipelines

including single-cell RNA sequencing (scRNA-seq)

DNA methylation

chromatin accessibility

Top Data Types

Image

Text

Document

Top Task Types

Computer Programming/Coding

Function Calling

Prompt + Response Writing (SFT)

RLHF

Segmentation

Company Overview

Company Overview **Mission** Immunome AI Biotechnologies is dedicated to decoding human immune memory using advanced machine learning foundation models. Our mission is to seamlessly bridge high-dimensional physical assays with predictive AI to map biological mechanisms, predict treatment responses, and accelerate targeted therapeutic discovery. **Services** We operate as a technical Data-as-a-Service (DaaS) partner, providing end-to-end multi-omic computational pipelines. Our core services include high-dimensional immune profiling, spectral cytometry panel design, and the characterization of therapeutic non-responders for complex clinical cohorts and pharmaceutical drug programs. **Unique Tools & Methods** Our proprietary infrastructure centers on the iMEMORY platform, a foundation model trained on over 160,000 biological graphs. We utilize Heterogeneous Graph Attention Networks (HetGAT) to integrate six distinct data modalities—including scRNA-seq, DNA methylation, and proteomics—into unified patient "immune endotypes." Our models are trained using highly optimized GPU-accelerated pipelines built on PyTorch Geometric, RAPIDS, and custom LMDB graph databases. **Specialized Industries** We operate exclusively at the intersection of Computational Biology and Artificial Intelligence, with deep domain expertise in immunology, transplant medicine, cardiovascular disease, and autoimmune therapeutic development. **Locations** Immunome AI Biotechnologies is headquartered in New York, NY, supporting research institutions and pharmaceutical partners globally. **History** Founded in January 2024, the company was built from the ground up to address the critical infrastructure bottleneck in multi-omic data integration. Evolving from an early platform concept into a comprehensive AI biotechnology firm, the company rapidly advanced to a seed-stage valuation of $30M by successfully deploying production-grade AI systems for major academic and commercial institutions. **Security & Infrastructure** Data integrity and security are foundational to our operations. We architect and maintain isolated, large-scale scientific computing environments using AWS EC2 instances equipped with multiple A100 GPUs and high-speed NVMe storage, ensuring complex patient data is processed securely and efficiently. **Workforce** Our operational model is lean and highly specialized. Led by a Founder and Principal AI Architect, the firm dedicates roughly 90% of its operational bandwidth directly to hands-on engineering, pipeline architecture, and biomedical research, ensuring partners interact exclusively with senior technical talent. **Achievements & Notable Clients** We provide active computational infrastructure and diagnostic insights for top-tier scientific institutions. Notable partners include the Translational Transplant Research Center at the Icahn School of Medicine at Mount Sinai, NYU Langone Center for the Prevention of Cardiovascular Disease and a Tier 1 Pharma, alongside leading tier-1 pharmaceutical drug programs utilizing our platform for lupus nephritis and kidney transplant research.

Security

Security Overview

Security & Privacy Overview Immunome AI Biotechnologies maintains rigorous, enterprise-grade security and privacy protocols tailored for highly sensitive biological and clinical data. Our security framework ensures the absolute integrity and confidentiality of all partner datasets, proprietary models, and AI training materials. **Clinical Data & Privacy Compliance:** All computational workflows strictly adhere to institutional regulatory standards. Extensive experience managing IRB protocols and multi-site clinical cohorts ensures that all patient-level data—including multi-omic profiles and flow cytometry datasets—is meticulously de-identified and anonymized prior to database ingestion or algorithmic training. **Secure Cloud Infrastructure:** All heterogeneous graph neural network (HetGAT) training and data processing occur on fully isolated, secure AWS EC2 environments. All data at rest across our 15 TB NVMe storage arrays is heavily encrypted. Network traffic and web-platform endpoints are secured and monitored via Cloudflare edge routing to prevent unauthorized access. **Endpoint & Access Management:** Operating as a lean, boutique firm allows for a zero-trust architecture with zero internal data fragmentation. Access to computing clusters, client datasets, and LMDB graph databases is tightly controlled through strict Identity and Access Management (IAM) roles, mandatory Multi-Factor Authentication (MFA), and secure SSH key protocols. **Data Isolation & Retention:** Client and project data are inherently siloed. We utilize dedicated cloud containers for specific data-as-a-service (DaaS) contracts, ensuring no cross-contamination between pharmaceutical partnerships, academic projects, or enterprise AI training tasks. Upon project completion, all proprietary data is purged according to strict, client-defined retention schedules.

Labeling Experience

Biophysical Simulation & Nonlinear Diffusion Modeling

Internal/Proprietary Tooling

Computer Code Programming

Computer Programming/Coding

Data Collection

Developed advanced mathematical simulations of oxygen and volatile anesthetic (isoflurane) diffusion from the vasculature to surrounding cellular structures. This involved engineering a nonlinear diffusion model of the neurovascular unit to analyze anesthetic delivery kinetics. The computational workflow required applying linear regression and complex parameter optimization to fit real-world in vivo data to Gaussian distributions. These custom biophysical models were rigorously validated against experimental data to test hypotheses regarding hypoxia-induced neural activity and delivery kinetics, culminating in first-author scientific publication.

2019 - Present

Foundation Model Graph Construction for Immune System Analysis (iMEMORY)

Internal/Proprietary Tooling

Text

Classification

Segmentation

Engineered, structured, and annotated a massive-scale multi-omic database consisting of 164,331 biological graphs to train iMEMORY, a proprietary heterogeneous graph neural network (HetGAT) foundation model. The complex data curation process required programmatically integrating six distinct biological data modalities—single-cell RNA, DNA methylation, chromatin accessibility, microRNA, proteomics, and genetic variants—into a unified architectural format. Custom GPU-accelerated processing and annotation pipelines were deployed on an AWS EC2 instance equipped with 8xA100 GPUs and 15 TB of NVMe storage to map the resulting biological graphs into an optimized LMDB format. This highly structured data allowed the model to successfully learn patient-level "immune endotypes" to predict clinical trajectories. To establish ground truth and verify pipeline accuracy, model predictions were rigorously validated utilizing specific clinical data cohorts from the Icahn School of Medicine at Mount Sinai, successfully characterizing mechanisms of treatment failure for tier-1 pharmaceutical therapeutic programs.

2024 - Present

High Dimensional Clinical Immune Phenotyping & Biomarker Curation

Internal/Proprietary Tooling

Medical Dicom

Segmentation

Classification

Engineered and executed specialized data processing and annotation pipelines for multi-parameter flow cytometry datasets across over 1,000 clinical samples to characterize the immune and platelet landscape in cardiovascular disease cohorts. This project required extracting structured, high-dimensional phenotypic signatures from raw biological outputs. Advanced data labeling techniques, including dimensionality reduction (t-SNE, UMAP) and complex spectral unmixing, were applied to manually gate, classify, and extract specific immune-cell clusters. These highly curated cellular datasets were then correlated with qualitative patient interviews and clinical outcomes to establish novel biomarkers and validate assay sensitivity across multi-site clinical cohorts. The resulting structured datasets directly supported advanced cardiovascular and platelet biology research.

2020 - Present