Zaid Yusuf

Curriculum vitae

Zaid Yusuf

Explore either tailored version below, or open and download the original PDF.

01 Bio-AI Research EngineerScientific machine learning, protein AI, and biological modeling

ML Research Engineer | Protein AI & Scientific ML

zaidyusuf.com · github.com/Yusufteppei · linkedin.com/in/zaid-yusuf-de · youtube.com/@theyusufzaid · yusufteppei11@gmail.com · Open to remote or local roles

Open PDF ↗Download PDF ↓

Summary

ML research engineer with ~4 years of experience across ML engineering, platform infrastructure and full-stack development, focused on scientific ML and protein AI. I implement research architectures from first principles—including diffusion over SE(3)/SO(3), AlphaFold2-style structure modeling, and single-cell foundation models—and bring the engineering background to make research code reproducible and deployable. 2x IMO contestant. Certified Kubernetes Administrator.

Professional Experience

Machine Learning Engineer

Digprom (Contract, part-time) · Jun 2023 – Present

  • Built and maintained machine-learning workflows and supporting data pipelines for government and university client projects at an IT consulting firm.
  • Developed backend software supporting data processing and ML applications.
  • Applied software engineering practices to improve reliability, reproducibility and maintainability.
  • Worked with engineering and domain teams to translate requirements into technical systems.

Full-Stack Engineer

Turing (Contract) · Jun 2024 – May 2025

  • Built and maintained backend systems and APIs supporting AI training projects.
  • Worked with cloud infrastructure and production deployment workflows, mostly Docker-based.
  • Designed reliable, data-driven application architectures.

Platform & Machine Learning Engineer

9ijakids (Contract) · Dec 2023 – Jul 2024

  • Built cloud-native infrastructure supporting scalable machine-learning systems.
  • Developed production ML solutions and inference workflows; automated deployment and experimentation pipelines.
  • Improved reliability, reproducibility and monitoring: adding monitoring cut average incident response time to one-fifth of its previous level and made it clear which team owned each incident.

Founding Engineer

Edulethics Africa (Contract) · Jan 2023 – Dec 2023

  • Led engineering strategy, software architecture and technical decisions across product development.
  • Designed and implemented the platform's core architecture, serving over 45,000 users.

Research Projects

ProtDiffusion: protein structure diffusion ↗

PyTorch, geometric deep learning

  • Research-oriented diffusion model over protein backbone geometry, started as a from-scratch reconstruction of RFdiffusion's ideas rather than a reproduction of it.
  • Represented structure as SE(3) rigid frames; implemented rotation and translation diffusion, SO(3) machinery, score-based modeling and Invariant Point Attention-inspired components.
  • Worked through torsion/dihedral geometry, atom placement, noise schedules, training and sampling; investigated numerical stability and training behavior.

AlphaFold2-mini ↗

PyTorch

  • Implemented core AlphaFold2 components from first principles: Evoformer, MSA row/column attention, outer product mean, triangle attention and multiplication, Invariant Point Attention and the structure module.

scGPT-mini: single-cell foundation model ↗

PyTorch, transformers

  • Implemented a transformer-based foundation model for single-cell gene expression: gene and expression embeddings, masked gene modeling, transformer encoder, data processing and training pipeline.

Breast cancer prognosis from TCGA data ↗

Python, scikit-learn, RNA-seq

  • Built an end-to-end workflow on transcriptomic and clinical data: RNA-seq processing, gene-expression feature engineering, clinical data integration, model development and evaluation.
  • Caught and corrected a data leakage issue that was inflating model accuracy.

Technical Writing

“A working intuition for diffusion on SO(3)” (Sep 2026) and “My Top Algorithms Across Transformer Architectures” (Oct 2026), published on zaidyusuf.com/writing.

Technical Skills

Machine learning
PyTorch, JAX, Hugging Face, transformers, diffusion models, geometric deep learning, LLM fine-tuning (SFT/LoRA), model evaluation
Computational biology
Protein structure prediction and generation, protein language models, single-cell transcriptomics, RNA-seq, cancer genomics, bioinformatics
Engineering
Python, C, C++, TypeScript, SQL, Bash; Docker, Kubernetes, AWS, Linux, CI/CD, PostgreSQL, FastAPI, Django

Education & Honors

  • Nile University of Nigeria, Bachelor's coursework in Computer Science | Sep 2017 – Jan 2020
  • International Mathematical Olympiad (IMO): 2x contestant (2015, 2016)
  • Pan-African Mathematics Olympiad (PAMO) 2015: Bronze medal
  • Cowbell Mathematics Competition 2014: 2nd place nationally (silver)
  • Certified Kubernetes Administrator (CKA)
  • Languages: English; German (B2)
02 ML Platform & AI InfrastructureProduction ML systems, infrastructure, and AI engineering

ML Platform & Infrastructure Engineer | LLM Inference & Serving | CKA

zaidyusuf.com · github.com/Yusufteppei · linkedin.com/in/zaid-yusuf-de · youtube.com/@theyusufzaid · yusufteppei11@gmail.com · Open to remote or local roles

Open PDF ↗Download PDF ↓

Summary

ML platform and infrastructure engineer with ~4 years of experience building cloud-native systems for machine learning, backend services and deployment pipelines. Certified Kubernetes Administrator with in-depth study of LLM inference and serving (KV caching, batching, parallelism, GPU capacity planning, SLO-driven observability) and hands-on research implementation experience in PyTorch and JAX. Strong systems fundamentals: built a relational database engine from scratch in C. 2x IMO contestant.

Professional Experience

Machine Learning Engineer

Digprom (Contract, part-time) · Jun 2023 – Present

  • Built and maintained machine-learning workflows and supporting data pipelines for government and university client projects at an IT consulting firm.
  • Developed backend systems supporting data processing and ML applications.
  • Applied software engineering practices to improve reliability, reproducibility and maintainability.
  • Worked with engineering and domain teams to translate requirements into technical systems.

Full-Stack Engineer

Turing (Contract) · Jun 2024 – May 2025

  • Built and maintained backend systems and APIs supporting AI training projects.
  • Worked with cloud infrastructure and production deployment workflows, mostly Docker-based.
  • Designed reliable application architectures and data-driven services.

Platform & Machine Learning Engineer

9ijakids (Contract) · Dec 2023 – Jul 2024

  • Built cloud-native infrastructure supporting scalable machine-learning systems.
  • Developed production ML solutions and inference workflows.
  • Automated deployment and experimentation pipelines.
  • Improved reliability, reproducibility and monitoring: adding monitoring cut average incident response time to one-fifth of its previous level and made it clear which team owned each incident.

Founding Engineer

Edulethics Africa (Contract) · Jan 2023 – Dec 2023

  • Led engineering strategy, software architecture and technical decisions across product development.
  • Designed and implemented the platform's core architecture, serving over 45,000 users.

Selected Projects

LLM Inference & Serving Study ↗

Systems design and implementation

  • Implementation-led study of the systems choices behind reliable LLM applications: prefill/decode architecture, KV-cache management, continuous and paged batching, prefix caching, speculative decoding, quantization, tensor and pipeline parallelism, GPU/VRAM capacity planning and model routing.
  • Covers serving metrics and observability: TTFT, TPOT, tokens/sec, p50/p95/p99 latency, KV-cache utilization, health checks, failure modes and SLO-driven scaling.

FensterSQL: relational database engine ↗

C, Flex/Bison

  • Built from scratch: SQL parsing, DDL/DML support, page-based storage, B+ tree indexing, catalogs and persistent metadata, disk-backed binary I/O, SQL shell, with manual memory management.

BioAgent: tool-calling agent ↗

LLM fine-tuning, SFT/LoRA

  • Fine-tuned open-source LLMs with SFT/LoRA to build a biology-focused tool-calling agent.
  • Studying agent reliability: tool boundaries, idempotency, retries and failure handling.

Research implementations in PyTorch ↗

Protein AI and scientific ML

  • ProtDiffusion (diffusion over SE(3)/SO(3) protein backbone geometry), AlphaFold2-mini (Evoformer, triangle attention, Invariant Point Attention) and scGPT-mini (single-cell foundation model), each implemented from first principles.

Technical Skills

Infrastructure
Kubernetes (CKA), Docker, AWS, Linux, CI/CD, PostgreSQL, cloud infrastructure
ML systems
LLM serving, KV caches, continuous/paged batching, prefix caching, speculative decoding, quantization, distributed inference, GPU/VRAM planning, observability and SLOs
ML and languages
PyTorch, JAX, Hugging Face, Python, C, C++, TypeScript, SQL, Bash
Backend and web
FastAPI, Django, Next.js, React

Education & Honors

  • Nile University of Nigeria, Bachelor's coursework in Computer Science | Sep 2017 – Jan 2020
  • International Mathematical Olympiad (IMO): 2x contestant (2015, 2016)
  • Pan-African Mathematics Olympiad (PAMO) 2015: Bronze medal
  • Cowbell Mathematics Competition 2014: 2nd place nationally (silver)
  • Certified Kubernetes Administrator (CKA)
  • Languages: English; German (B2)
Get in touch ↗ GitHub ↗ YouTube ↗

Private login