Browse · People
Person · Latent Space wiki

RJ Honicky

MiraOmics co-founder and CTO working on AI for biological data analysis, and co-host of Latent Space’s AI for Science podcast.

RJ Honicky is co-founder and CTO of MiraOmics, where he builds the engineering and data-science capabilities behind its analytics and AI/ML platform. His work includes leadership of MiraTyper, an automated cell-type annotation project. He also co-hosts Latent Space’s AI for Science podcast with Brandon Anderson.[2][5][1]

MiraTyper and biological analysis

MiraOmics co-founder Eugene Bolotin credited Honicky with “driving this project” in an announcement of MiraTyper’s availability on LatchBio. MiraTyper predicts cell types from single-cell RNA-sequencing data. Bolotin separately credited Pythia Biosciences with providing training data.[5]

Honicky treats model uncertainty as a possible clue to biology. In a February 4, 2026 post, he reported that low-confidence MiraTyper predictions sometimes reflect unusual biological conditions, such as stress states. His explanation was that the transcriptomic data fall outside the model’s training distribution. He kept alternative explanations attached: missing training data and batch effects can also produce uncertainty. These were his observations, rather than a validated diagnostic rule.[6]

Honicky proposed a workshop combining cell annotation with agent-assisted analysis. In a workshop invitation, he proposed analyzing a cervical-cancer slide using Claude Code, MiraTyper and shared SKILL.md files, with alternative tools welcome. He described cell typing as “the join key of biology” and emphasized analysis participants could understand and validate, alongside visualization, reporting, publishing and tuning agents to their workflows.[5]

AI for Science podcast

Latent Space introduced Honicky and Brandon Anderson as the hosts of its new AI for Science podcast on January 28, 2026, in an episode featuring Andrew White. The publisher described Honicky’s MiraOmics work as building AI models and services for single-cell data, spatial transcriptomics and pathology-slide analysis. Reacting afterward to White’s chemistry reward-hacking story, Honicky asked “Reward hacking or insight?”—suggesting that apparently misguided model outputs might warrant investigation, even when particular proposals are infeasible.[1][6]

Earlier engineering career and omics work

Before his MiraOmics work, Honicky’s engineering responsibilities spanned telecommunications, mobile computing and enterprise storage. He was Director of Engineering at Tarana Wireless, overseeing AI/ML, data science and manufacturing test systems, following senior technical and research positions at Nokia Research and Network Appliance. This earlier experience supplies context for his present responsibility for data infrastructure and engineering at MiraOmics.[2]

By May 2025, Honicky was publicly associated with AI applications in omics. Pythia advertised him as Senior AI Consultant at miraomics for a May 29 webinar with Eugene Rakhmatulin. Its planned topics included LLM-assisted cell typing, mechanism-of-action hypothesis generation, pathway interpretation and interactive questions about omics data, along with retrieval-augmented generation and fine-tuning.[4]

Timeline

  1. role

    Advertised as MiraOmics Senior AI Consultant for an omics webinar
    By May 2025, Honicky was publicly associated with AI applications in omics. Pythia advertised him as Senior AI Consultant at miraomics for a May 29 webinar with Eugene Rakhmatulin.[4]

  2. release

    Introduced as AI for Science co-host alongside Brandon Anderson
    Latent Space introduced Honicky and Brandon Anderson as the hosts of its new AI for Science podcast on January 28, 2026, in an episode featuring Andrew White.[1]

  3. research

    Suggests MiraTyper uncertainty may reveal unusual biology
    In a February 4, 2026 post, he reported that low-confidence MiraTyper predictions sometimes reflect unusual biological conditions, such as stress states. His explanation was that the transcriptomic data fall outside the model’s training distribution. He kept alternative explanations attached: missing training data and batch effects can also produce uncertainty. These were his observations, rather than a validated diagnostic rule.[6]

From the Latent Space archive

Original posts and separately reviewed explanations.

Sources

  1. 1
  2. 2
    About Us – MiraomicsAccessed 2026-09-20
  3. 3
  4. 4
  5. 5
    MiraOmics | LinkedInAccessed 2026-09-20
  6. 6

Model source and teaching reviews passed. Human editorial review pending. This profile describes the sources as of 2026-09-20.

Latent Space

Join Latent Space.

Sign in with Google, or get the newsletter by email.

Loading Google sign-in…

or subscribe by email

Free. Unsubscribe any time.