AI engineering & computational biology

Viktor Petukhov, PhD

I build production AI systems, research software and full-stack products — from defining the problem to delivering a working solution.
  • Production LLM Systems
  • RAG & Agents
  • Bayesian Statistics
  • Machine Learning
  • Data Visualisation
  • Computational Biology
  • Python
  • Julia
  • R
  • C++
  • TypeScript
  • 6,800+ citations · November 2026
Viktor Petukhov
I hold a PhD in Biostatistics and Bioinformatics from the University of Copenhagen, with research at Harvard Medical School. My work spans Bayesian methods and open-source tools used by thousands of scientists, production LLM systems and full-stack products. As an independent AI and biotech consultant, former CTO and founder, I combine hands-on engineering with scientific methods and product strategy.
About me

How I can help

From evaluating an idea to building and operating the system, with experience across industry, biotech and non-profits.
AI systems & products
  • Production LLM systems
    Build document intelligence, RAG, hybrid search and agent workflows with source verification and model evaluation. Experience with Pydantic AI, LiteLLM, PostgreSQL / pgvector and EU-hosted deployments.
  • Full-stack products & prototypes
    Turn a problem into a working product: data collection, AI pipelines, APIs, search and user interfaces. Python, FastAPI, TypeScript, React / Next.js, Docker and GCP.
  • Document & workflow automation
    Process scans and office documents with OCR and LLMs, and connect workflows to Microsoft 365, Teams, SharePoint and ERP systems.
  • Technical strategy & leadership
    Assess feasibility, scope an MVP, define architecture and hire an engineering team. Experience as a start-up CTO, founder and independent consultant, including technical due diligence for biotech investment.
Computational biology & modelling
  • Single-cell & spatial transcriptomics
    Analyse scRNA-seq and spatial data, develop quality-control metrics, and interpret cell composition and gene-expression changes. Experience with 10x Xenium, MERFISH, Scanpy and Seurat.
  • Multi-omics pipelines
    Build automated analysis workflows for scRNA-seq, CITE-seq, TCR/BCR-seq and ATAC-seq, including work supporting CAR-T cell-therapy development.
  • Statistical methods & research software
    Develop Bayesian models, graph methods and scientific tools in R, Julia, C++ and Python. Lead-author software includes Baysor, dropEst and cacoa; co-first-author work includes conos.
  • Benchmarking & independent evaluation
    Evaluate algorithms and biological results, quantify the effects of processing errors, and build reproducible benchmarks for scientific and AI systems.

Selected projects

Production AI, independent products and scientific software. Client work under NDA is described without naming the client.
Article Visual Search
  • Sole developer
  • NLP
  • Open source
Independently built an application for exploring PubMed and Semantic Scholar literature on an interactive map. Combines embeddings, clustering and optional LLM-generated cluster names and summaries. Developed intermittently during 2023–2026.
Steps to get started
1
Understand the problem
We clarify your goals, the available data and resources, and how we will judge success.
2
Choose an approach
We assess feasible options, agree on scope and choose the tools and evaluation methods.
3
Agree on a plan
I propose concrete deliverables, milestones and timelines.
4
Build & review
I deliver incrementally, usually with visible progress each week, and adjust the work using your feedback.
Let’s discuss your project
Get in touch about AI engineering, computational biology or research collaborations. Early ideas are welcome too.