Applied AI & Automation Engineer building intelligent systems that combine machine learning, domain knowledge, and automation to solve real-world problems.
I design and deploy AI systems across scientific workflows, predictive modelling, retrieval systems, and autonomous agents. My work focuses on turning complex processes into reliable AI-powered tools that reduce manual effort and improve decision-making.
Currently working as a Senior Microbiology Analyst while building production-grade AI systems in my own projects.
- Applied Machine Learning Systems
- Autonomous AI Agents & Workflow Automation
- Retrieval Augmented Generation (RAG)
- Predictive Modelling & Simulation
- Domain-Specific AI Systems
- Scientific AI & Laboratory Automation
Hybrid AI system for phenotype-based bacterial identification.
- Combines deterministic microbiology logic, machine learning, and LLM reasoning
- 8,700 curated laboratory records across 140 bacterial genera
- Achieved 95.1% genus classification accuracy
- Full-stack architecture with Flask API and React frontend
Tech: Python, XGBoost, FLAN-T5, BART, Flask, React, Hugging Face
Demo: https://huggingface.co/spaces/EphAsad/BactAID-Demo
Predictive food microbiology simulation platform for modelling Listeria monocytogenes growth and shelf-life risk.
- Machine learning prediction of microbial growth curves
- Monte Carlo simulation for uncertainty estimation
- Sensitivity analysis to identify dominant environmental risk drivers
- FastAPI backend with React frontend interface
Tech: Python, CatBoost, FastAPI, React, Monte Carlo Simulation, Predictive Modelling, Ollama
Privacy-first local autonomous coding and workflow agent.
- Planner/executor architecture separating reasoning from execution
- Tool orchestration system with 60+ integrated tools
- Schema validation and error recovery for reliable task execution
- Designed for fully local AI development using Ollama
Tech: Python, Ollama, Multi-Agent Systems, Tool Routing, JSON Schema
Domain-adaptive embedding model for multi-domain retrieval systems.
- LoRA-based architecture enabling modular domain adapters
- Reinforcement learning policy for automatic domain routing
- Trained on NLI, semantic similarity, and paraphrase datasets
- Achieved 2.7× improvement over baseline domain classification
Tech: PyTorch, Transformers, LoRA, Contrastive Learning, Reinforcement Learning
Model: https://huggingface.co/EphAsad/DomainEmbedder
AI-powered laboratory SOP retrieval assistant.
- Hybrid RAG system combining FAISS vector search and BM25 keyword retrieval
- Fine-tuned domain embeddings for laboratory terminology
- Provides source-grounded procedural guidance for regulated environments
Tech: Python, FAISS, BM25, Streamlit, Ollama
Laboratory workflow automation platform built using Access VBA.
- Parent-child database synchronization across 20+ laboratory test workflows
- Automated colony count parsing and result aggregation
- Media batch traceability and regulatory compliance tracking
- Backup and recovery automation for laboratory data integrity
Tech: VBA, Microsoft Access, SQL, Laboratory Automation
Python, JavaScript, SQL, VBA
XGBoost, Random Forest, Transformers, Transfer Learning, LoRA, Model Evaluation
RAG, LLM Orchestration, Domain-Specific Embeddings, Prompt Engineering
PyTorch, Scikit-learn, Pandas, NumPy, FAISS, BM25
FastAPI, Flask, React, Streamlit, Hugging Face Spaces
Git, GitHub Actions, Local LLMs (Ollama)
Clinical Microbiology, Bacterial Identification, Phenotypic Analysis, ISO 17025 Laboratory Workflows
BactAI-D: A Hybrid, Confidence-Aware AI System for Phenotype-Based Bacterial Identification
Zenodo Preprint
https://doi.org/10.5281/zenodo.18089381
DomainEmbedder v2.6: Domain-Adaptive Embedding Model for Cross-Domain Retrieval
Hugging Face Model Repository
Decomposing Sex Differences in Mortality Across Age and Cause in England and Wales, 1915–2015
SocArXiv Preprint
LinkedIn
https://linkedin.com/in/zain-asad-1998eph
GitHub
https://github.com/EphraimAsad
Hugging Face
https://huggingface.co/EphAsad
Email
zainasad98@gmail.com
- Autonomous AI agents
- AI-powered scientific tooling
- Predictive modelling systems
- Local-first AI infrastructure
- AI-driven workflow automation
Always open to interesting technical discussions and collaborations.
