cv
Basics
| Name | Rebonto Haque |
| Label | PhD Researcher (Statistics) — Interpretable ML for protein sequence–structure–function (therapeutic antibodies) |
| rebonto.haque@magd.ox.ac.uk | |
| Phone | +447404265414 |
| Summary | PhD researcher working on interpretable deep learning for protein sequence–structure–function relationships utilising SAE-driven mechanistic interpretability studies. Experienced in structure-conditioned generative models (ESM-IF and finetuned variants), high-dimensional biological data, and HPC-based ML workflows. |
Work
-
2024.09 - Present Master’s + PhD Student Researcher
Oxford Protein Informatics Group, University of Oxford
Research under Prof. Charlotte M. Deane on interpretable deep learning for protein sequence–structure–function relationships, with applications to antibody–antigen interactions and cross-reactivity.
- Investigating correlations between sequence likelihood of structure-conditioned inverse folding models and cross-reactivity
- Exploring conditioning approaches for ESM-IF to maintain bi-specificity in generated sequences
- Trained Sparse Autoencoders (SAEs) for mechanistic interpretability and inference-time steering of antibody language models (IgGen, ProGen2, p-IgGen)
- Identified latents associated with antibody-specific concepts (e.g., V/J germline identity, CDR identity)
- Used SAE latents to steer generation toward germline-specific sequences while maintaining diversity, novelty, and antibody-likeness
- SAE paper accepted to MLSB workshop (co-located at NeurIPS 2025)
-
2024.07 - Present ML Intern / ML Software Developer (Part-time)
TraitSeq
Part-time ML work at an agritech company using machine learning to enhance breeding strategies.
- Analysed high-dimensional metabolomics, transcriptomics, and genomics data to inform model predictions
- Implemented differential expression (edgeR) for feature selection to increase predictive accuracy
- Integrated edgeR + ML feature importance + gene regulatory network connectivity to identify biomarkers
- Investigated whether stressed phenotypes can be predicted using non-stressed expression data
- Developed ensemble learning strategies improving model generalisation by ~15%
-
2023.06 - 2023.08 Research Intern
Earlham Institute
Developed an end-to-end image analysis and ML pipeline for biological specimens (wildflower seeds).
- Selected among top 14 students from 500+ applicants
- Built segmentation → feature extraction → classification pipeline; designed ImageJ plugins for automated processing
- Analysed 32,915 seeds across 20 species
- Achieved 98% balanced accuracy (improving from ~90%)
- Reduced runtime from ~20 hours to ~2.5 hours
- Work submitted for publication (under review)
-
2022.08 - 2022.09 Intern
International Centre for Diarrhoeal Disease Research, Bangladesh
Laboratory training and shadowing across sample handling, toxicology preparation, and immunology workflows.
- Sample handling: labelling, processing, storage; biosafety procedures
- Prepared samples for Atomic Absorption Spectrometry and Ficoll-Paque PBMC separation
- Exposure to ELISA, flow cytometry immune cell profiling, serum bactericidal assays, and microbiological analysis
Extracurriculars
-
2025.12 - Present Oxford, UK
Welfare Officer
Magdalen MCR, University of Oxford
Welfare Officer in the Magdalen Middle Common Room, organising welfare events and peer support activities.
- Organising regular welfare events and collaborating with the welfare team to ensure adequate student support
- Serving as a student mentor to ensure welfare provisions are accessible to the student body
- Currently undergoing Peer Support Training
-
2023.03 - 2023.04 Oxford, UK
Student Consultant
The Oxford Strategy Challenge
Developed a strategy action plan for biotech startup TrialeX to increase participation of under-represented groups in clinical trials.
- Designed targeted marketing strategies using data-driven insights to tailor outreach
- Proposed strategies to increase participation among homeless populations
- Presented findings to founders alongside competing teams
-
2023.03 - 2025.04 Oxford, UK
President / Secretary
Oxford University Chemistry and Biochemistry Society
Led society operations and growth; previously served as Secretary before being elected President (March 2024).
- Managed society funds with Treasurer to support sustainable growth and budgeting
- Oversaw a 1,000+ member mailing list to drive engagement and event participation
- Built collaborations with other student groups; delivered large-scale events with ~150 attendees
-
2023.02 - 2024.01 Oxford, UK
Outreach Officer
Oxford University Bangladesh Society - University of Oxford, UK
Led outreach and admissions support activities for prospective Bangladeshi students.
- Provided admissions support to international Bangladeshi students
- Offered advice on personal statements, interview preparation, and course selection
- Conducted live webinars followed by Q&A sessions with turnouts of ~40 students
- Collaborated with an educational agency to provide advice on the visa process
- Mentored applicants individually; one mentee was accepted to the University of Oxford
-
2022.10 - 2023.08 Oxford, UK
Academic Affairs Representative
St. Peter's College, University of Oxford
Served as the academic representative for the college, liaising with senior tutors and college committees.
- Served as the academic voice for ~350 students, liaising with the college Master, Head Tutor, and senior academics through bi-termly Academic Affairs Committee meetings
- Led a comprehensive feedback initiative, gathering subject-specific student perspectives on teaching quality and concerns
- Synthesised findings into detailed reports and presented them to the committee, driving constructive dialogue on improving teaching provision and resolving identified issues
-
2022.10 - 2023.08 Oxford, UK
2nd Year Biochemistry Student Representative
Oxford Biochemistry Department
Represented the academic interests of second-year biochemistry students to senior faculty.
- Represented the academic interests of 110 second-year biochemists to senior faculty in the Joint Consultative Committee, Steering Committee, and Undergraduate Teaching Committee
- Implemented systematic feedback collection via surveys, group discussions, and in-person outreach, increasing student response rates by 75% compared to other year groups
- Advocated for and secured the installation of desks in the new teaching centre, addressing widespread student concerns despite initial resistance from faculty
- Persuaded the teaching committee to provide additional exam preparation resources, improving academic support for students
Education
-
2025.01 - 2029.01 Oxford, UK
PhD
University of Oxford — Magdalen College
Statistics
- Interpretable machine learning for protein sequence–structure–function relationships in therapeutic antibodies
-
2021.01 - 2025.01 Oxford, UK
-
2009.01 - 2021.01 Bangladesh
Secondary Education
S.F.X. Greenherald International School
- A levels: 5 A* (Biology, Chemistry, Physics, Mathematics, Computer Science)
- O levels: 8 A*
Awards
- 2023.01.01
OxBioHackathon 2023 — Winner (First Place)
Oxford University Biotechnology Society
Won first place for a project pitch exploring therapeutic potential of novel piRNA technology, addressing limitations in existing gene therapy approaches.
- 2019
Cambridge Highest in the World Academic Achievement Award (O Level Physics)
Cambridge Assessment International Education
- 2019
Cambridge Highest in the Country Academic Achievement Award (O Level Bangladesh Studies)
Cambridge Assessment International Education
Publications
-
2025.12.01 Mechanistic Interpretability of Antibody Language Models Using SAEs
arXiv (preprint)
First-author preprint on SAE-driven mechanistic interpretability methods for antibody language models.
-
Integrating Machine Learning, Deep Learning, and Image Analysis for Seed Species Classification
In review
Manuscript describing an end-to-end image analysis + ML pipeline for wildflower seed species classification.
Skills
| Computational & Machine Learning | |
| PyTorch | |
| TensorFlow | |
| Deep generative models | |
| Representation learning | |
| Interpretability | |
| Statistical modelling | |
| Differential expression (edgeR) |
| Biological Data | |
| Protein sequence & structural data | |
| Antibody annotation (ANARCI, OAS, PLAbDab) | |
| Transcriptomics | |
| Genomics | |
| Metabolomics |
| Systems | |
| Linux | |
| HPC (SLURM) | |
| GPU computing | |
| Python | |
| Java |
Projects
- 2024.09 - Present
SAE-driven mechanistic interpretability for antibody language models
Sparse Autoencoder (SAE) training and latent analysis for steering antibody sequence generation while preserving diversity and antibody-likeness.
- Latent discovery for V/J germline identity and CDR identity
- Inference-time steering for germline-specific generation
- 2023.06 - 2023.08
Wildflower seed image analysis + classification pipeline
End-to-end pipeline integrating segmentation, feature extraction, and classification for seed species identification.
- 32,915 samples across 20 species
- 98% balanced accuracy
- Runtime reduced from ~20h to ~2.5h