Linfeng Wang

Machine Learning Engineer at Biographica
ML Engineer building deep learning models on genomic sequence

My Journey in AI and Healthcare

6+ years building ML models (MRes, PhD and industry)
Biochemistry-trained since 2016; building ML for genomics since 2019
10 Peer-reviewed publications, including 5 as first author
(2 further papers currently being peer reviewed)

Resume

I build deep learning models on genomic sequence: graph attention networks, LSTMs, and now gene expression prediction, from research prototype to cloud pipeline.

Experience

Machine Learning Engineer
Biographica
Apr 2026 to Present

Machine Learning Consultant
Deep Science Ventures
Mar 2025 to Oct 2025

Machine Learning Intern
LinkGevity
Aug 2024 to Oct 2024

Data Science Intern
ByteDance
Aug 2022 to Feb 2023

Trainee
Syngenta
Jun 2017 to Aug 2017

Education

PhD - Machine learning for Infectious Disease Genomics (group website)
London School of Hygiene and Tropical Medicine
Oct 2021 to Mar 2026

Master of Research Bioengineering (Merit)
Imperial College London
Oct 2019 to Oct 2020

BSc Biochemistry (First Class Honours)
King's College London
Sep 2016 to Jul 2019

Skills

PyTorch / JAX / TensorFlow / PyG / Hugging Face
Python / C++ / R / scikit-learn / NumPy / Pandas

AWS / GCP / Azure / Linux / Bash / Git / Nextflow
BWA-MEM / SAMtools / BCFtools / GATK / PLINK2 / BEAST2

Publications

2026

Wang L, Campino S, Clark TG, Phelan JE. Decoding positive selection in Mycobacterium tuberculosis with phylogeny-guided graph attention models. BMC Bioinformatics. doi:10.1186/s12859-026-06583-0

2025

Wang L, Thawong N, Thorpe J, Higgins M, Ik MTK, Sawaengdee W, Mahasirimongkol S, Perdigão J, Campino S, Clark TG, Phelan JE. TOAST: a novel tool for designing targeted gene amplicons and an optimised set of primers for high-throughput sequencing in tuberculosis genomic studies. BMC Genomics 26(1). doi:10.1186/s12864-025-12247-9
Wang L, Campino S, Clark TG, Phelan JE. LSTM-based deep learning model for the discovery of antimicrobial peptides targeting Mycobacterium tuberculosis. Bioinformatics Advances. doi:10.1093/bioadv/vbaf274
Wang L, Campino S, Clark TG, Phelan JE. A multi-stage machine learning framework for stepwise prediction of tuberculosis treatment outcomes: integrating gradient boosted decision trees and feature-level analysis for clinical decision support. Research Square (preprint). doi:10.21203/rs.3.rs-7558046/v1
Thawong N, Sriloha P, Phelan JE, Phornsiricharoenphant W, et al., Wang L, Clark TG. Genomic analysis of tuberculosis in Thailand. Research Square (submitted).
Data Study Group Team. Detecting shallow gas from marine seismic images. The Alan Turing Institute. Final report

2024

Wang L, Lim DR, Phelan J, Campino S, Hibberd ML. WGS of TB in the Philippines. Sci Rep 14:70471.
Phelan J, Niazi F, Wang L, Ngwana-Joseph GC, et al. TGV: tools to visualize transmission graphs. NAR Genomics Bioinform 6(4).

2023

Wang L, Campino S, Phelan J, Clark TG. Mixed infections in drug-resistant TB. Sci Rep 13:1–8.
Libiseller-Egger J, Wang L, Deelder W, Campino S, et al. TB-ML: comparing ML models for TB drug resistance. Bioinform Adv.

(More in Google Scholar…)

DOWNLOAD RESUME

Current work: Biographica

Gene expression from DNA sequence: deep learning for crop gene-editing targets

At Biographica I build deep learning models that predict gene expression from DNA sequence, which we use to prioritise gene-editing targets in crops. Alongside the models I work on a gene-identification pipeline that runs across tomato accessions, and we are extending the same approach to yeast.

Thesis

Infectious Disease Genomics Meets AI: Scalable Models, Interpretable Tools, and Data-Driven Diagnostics

Phylogeny-guided graph attention for genomic selection (TB)

LSTM generative model for antimicrobial peptides

TOAST: Custom Amplicon Designer for TB Diagnostics

Clinical Outcome Prediction using Machine Learning

Dissecting Mixed Infections with ML Models

Understanding TB transmission in the Philippines

EXPLORE ALL

Phylogeny-guided attention visualisation

TB transmission analysis in the Philippines

Fun Experience

Turing DSG Hackathon: Detecting Shallow Gas from Marine Seismic Images (Alan Turing Institute: Data Study Group challenge)

Over a week of intense collaboration with a diverse team of participants from all over the world, we:

Explored CNN architectures and ensemble approaches for geospatial pattern recognition.
Developed classification, segmentation, and object detection models for shallow gas pocket identification in noisy, low-contrast marine datasets.
Contributed to the end-of-sprint report for industrial partner BP, presenting results with interpretability-focused visualizations.

Group photo from the pathogen genomics course

Teaching Trips

Data & AI: for pathogen genomics (Philippines, Indonesia, Thailand, UK)

Week-long intensive courses taught at a range of locations to diverse groups of researchers from all over the world:

Teaching Bash programming and the use of command-line based software.
Teaching the mathematics and biology behind the genomics sequencing pipeline we developed in house.
Building a RAG chatbot with LangChain and ChromaDB, grounded in the course content, for interactive participant Q&A.
Individual problem-solving workshops: providing AI and bioinformatics solutions to real research problems on a case-by-case basis.