AI R&D Leader & Platform Architect

Tzu-Tang Lin

Building AI Platforms from Strategy to Scale

With 8+ years across platform strategy, solutions architecture, and hands-on AI delivery, I turn complex scientific and enterprise workflows into scalable agentic systems—with deep expertise in AI for Science.

Resume (PDF)
Portrait of Tzu-Tang Lin

About Me

I am an AI platform and solutions architecture professional with 8+ years of experience across academic research, startups, FinTech, and AI for Science. My work spans AI model development, scalable research platforms, high-throughput computing, AI tool integration, and end-to-end solution delivery.

As AI R&D Manager and Special Assistant to the General Manager at Repurgenesis, I lead AI platform strategy, R&D roadmapping, cross-functional execution, and product delivery. I have contributed to building both the company and its core AI platform from 0 to 1, integrating AI models, data pipelines, and scientific tools into practical R&D solutions. Alongside this role, I am pursuing an M.S. in Data Science at National Taiwan University.

Previously, I conducted AI research at Academia Sinica, the University of Florida, and Virginia Tech, developing predictive, generative, and multimodal models, biomedical AI platforms, and large-scale computational workflows. At Cathay Financial Holdings, I worked on enterprise data science and FinTech applications. These experiences enable me to bridge research depth with product and business requirements.

I have authored multiple peer-reviewed AI publications as first author and was selected as a 2026 NVIDIA GTC Poster Finalist for an AI orchestration framework for drug R&D. Going forward, I am focused on AI platform strategy, solutions architecture, AI product development, and technical leadership—turning advanced AI capabilities into scalable products with real-world and business impact.

Core Expertise
  • AI Platform Strategy & Solutions Architecture
  • Agentic AI, RAG & Knowledge Systems
  • AI R&D Leadership & Product Delivery
  • AI for Science & High-Throughput Computing

News

  • Apr 2026 — Co-authored Harnessing Sequence Embedding and Ensemble Learning to Identify Antifungal Peptides with Low Hemolytic Risk, published in ACS Omega.
  • Mar 2026 — Selected as a Finalist (Top 8/158) at NVIDIA GTC 2026 for an AI-orchestrated drug repurposing platform.
  • Mar 2026 — Earned four NVIDIA-Certified Associate credentials across AI infrastructure, multimodal generative AI, LLMs, and accelerated data science.
  • Sep 2025 — Joined Repurgenesis as AI R&D Manager & Special Assistant to the General Manager, leading AI platform strategy and product delivery.
  • Dec 2023 — Co-received the 20th National Innovation Award for AI Fleming, an AI-powered therapeutic peptide discovery platform.

Experience

AI R&D Manager & Special Assistant to the GM - Repurgenesis
Sep 2025 – Present
  • Lead AI platform strategy and R&D roadmapping, translating scientific workflows into prioritized modules, development milestones, and product plans.
  • Architected the integration of four domain-specific R&D modules into a unified agentic AI platform combining scientific tools, RAG, knowledge graph reasoning, automated SOPs, evidence scoring, and human-in-the-loop review.
  • Lead development of a scalable structure-based drug repurposing platform integrating proprietary data, automated molecular preparation, docking, affinity prediction, and candidate ranking.
  • Evaluated NVIDIA BioNeMo, NeMo Agent Toolkit, and NIM for AI for Science architecture and GPU-accelerated delivery; first-authored a 2026 NVIDIA GTC Poster Finalist submission ranked Top 8 of 158.
  • Support the General Manager across corporate strategy, fundraising preparation, R&D prioritization, external partnerships, and product positioning.
  • Translate technical capabilities into investor materials, executive presentations, board-level AI education, and partner demonstrations.
Research Assistant - University of Florida — College of Pharmacy
May 2024 – Aug 2025
  • Developed a multimodal foundation model integrating RNA sequences, molecular representations, and 3D structural information for RNA–ligand binding affinity prediction and virtual screening.
  • Built large-scale AI training, tuning, and screening workflows on a 100+ NVIDIA A100 GPU cluster, processing up to 10 million molecular inputs through automated data pipelines and SLURM-managed computing.
Research Assistant - Virginia Tech — Dept of Computer Science
Aug 2023 – Apr 2024
  • Developed PathoVF, an AI bioinformatics tool combining multimodal DNA representations, protein descriptors, protein language model embeddings, and machine learning for pathogenicity and virulence prediction.
  • Built TB-scale metagenomic pipelines for sequence analysis, taxonomic profiling, and detection of antimicrobial-resistance and virulence-associated signals.
Data Scientist - Cathay Financial Holdings — Data Science Lab
Dec 2022 – May 2023
  • Developed enterprise AI proof-of-concept solutions across FinTech and healthcare, including signature recognition, synthetic financial data generation, federated AI agents, and clinical risk prediction.
  • Built an EHR-based ICU sepsis early-warning model using a Feature Tokenizer Transformer, achieving 0.915 accuracy in internal evaluation.
  • Used AWS EC2, SageMaker, and S3 for scalable model development and secure data workflows; partnered with business, healthcare, and technical stakeholders and instructed Cathay General Hospital’s Data Science Workshop.
Bioinformatics Engineer / Research Assistant - Academia Sinica — Institute of Information Science
Jul 2018 – Dec 2022
  • Led development of AI Fleming, an AI-driven antimicrobial peptide discovery platform spanning data curation, predictive modeling, generative design, candidate prioritization, and experimental validation.
  • Developed CNN, LSTM, Doc2Vec, ensemble, and WGAN-GP models for peptide activity prediction and de novo sequence generation.
  • Designed AI-generated peptide candidates; 2 of 8 experimentally validated candidates demonstrated in vitro activity against cancer cells and antibiotic-resistant bacteria, with MIC values of 2–45 μg/mL.
  • Integrated multiple predictive models into the public AI4AXP platform and developed the PC6 protein encoding method used in AI4AMP; contributed to work recognized with the 20th National Innovation Award.

Education

Expected Jun 2027
M.S. in Data Science
National Taiwan University, College of Electrical Engineering and Computer Science
Graduate study in data science alongside full-time AI R&D leadership.
Sep 2015 – Jun 2019
B.S. in Agronomy (Specialization in Experimental Design and Biostatistics)
National Taiwan University
Awarded Undergraduate Research Fellowship, Ministry of Science and Technology (MOST), Taiwan.

Credentials

Mar 2026
AI Infrastructure and Operations
NVIDIA-Certified Associate
Mar 2026
Generative AI Multimodal
NVIDIA-Certified Associate
Mar 2026
Generative AI LLMs
NVIDIA-Certified Associate
Mar 2026
Accelerated Data Science
NVIDIA-Certified Associate

Selected Platforms

AI4AXP
Academia Sinica — Institute of Information Science
AI4AXP, a user-friendly web platform for predicting peptide activities (antibacterial, antifungal, anticancer, antivirus, hemolysis).
AI Fleming Introduction Video
Academia Sinica — Institute of Information Science
AI Fleming: a generative-plus-predictive AMP platform using GANs and AI4AMP to design, identify, and validate new candidates.

Selected Technical Projects

Cross-Domain Multimodal Contrastive Learning for RNA–Ligand Binding Affinity
RNA Ligand CLIP LLM Python Prediction
Cross-Domain Multimodal Contrastive Learning for RNA–Ligand Binding Affinity
CLIP-style cross-modal foundation model integrating RNA sequences, LLM SMILES (RiNALMo/SMI-TED), and 3D features for RNA–small-molecule affinity; supports downstream fine-tuning.
GIGN for RNA–Ligand Binding Affinity
RNA Ligand GNN GIGN Python Benchmarking Prediction
GIGN for RNA–Ligand Binding Affinity
Fine-tuned/retrained Geometric Interaction Graph Neural Network (GIGN) for RNA–ligand interaction modeling with packaged training/inference.
Sepsis Early Prediction with FT Transformer (CGH)
Healthcare EHR Transformer Python Pipeline Prediction
Sepsis Early Prediction with FT Transformer (CGH)
EHR-based early sepsis prediction prototype using FT Transformer—data processing → featurization → training/evaluation; reproducible notebooks & scripts.
RLaffinity — 3D-CNN Contrastive Inference
RNA Ligand 3D-CNN Contrastive Python Benchmarking Prediction
RLaffinity — 3D-CNN Contrastive Inference
Reproducible inference & retraining workflow for a 3D-CNN contrastive model on nucleic acid–ligand binding; scripts for data prep and evaluation.
Hariboss Processing & Pocket Extraction
RNA Ligand Bash Pipeline Dataset
Hariboss Processing & Pocket Extraction
Hariboss cleaning, RNA/ligand splitting, and pocket extraction; SLURM batch for HPC and integration with RNA3DB.
PDBbind RNA–Ligand Processing Pipeline
RNA Ligand Bash Python Pipeline Dataset
PDBbind RNA–Ligand Processing Pipeline
Batch download/parse from PDBbind NL, RNA chain & ligand extraction, format conversions, and sequence export for training/pocket maps.

Get in Touch

I welcome conversations about AI platform strategy, solutions architecture, agentic AI, and AI for Science.