Summer 2026

Machine Learning Intern

Posted on 3/25/2026

Insitro

Insitro

201-500 employees

ML-driven predictive drug discovery and development

Compensation Overview

$35 - $65/hr

San Bruno, CA, USA

Hybrid

Hybrid work schedule; remote option available based on team mentor location.

Bachelor's, Master's, PhD

Category
AI & Machine Learning (2)
,
Required Skills
LLM
Python
Machine Learning

Get referred to Insitro

See people who can refer or advise you

Requirements
  • Working towards a Bachelor of Science, Master of Science, or Doctor of Philosophy in engineering, computational biology, systems biology, computer science, mathematics, statistics, life science, chemistry, physics, or a related field.
  • Proficiency in one or more general-purpose programming languages; Python is primarily used.
  • Interest in using and developing brand new statistical and machine learning methods inspired by real problems.
  • Curiosity about human physiology or disease biology.
  • Committed to writing high-quality, well-commented code and documentation.
  • Ability to communicate effectively and collaborate with people of diverse backgrounds and job functions.
  • Passion for making a difference in the world.
Responsibilities
  • Partner directly with a team mentor to develop and/or apply machine learning methods to a process and analyze large scale datasets from multiple modalities over a Summer 2026 internship (11-12 weeks).
  • Work on areas including Computational Biology, Methods for Omics and Imaging data modalities, Statistical and Translational Genetics, Integrative Phenotyping, Molecular Machine Learning, and Computational Microscopy.
  • Develop, productionize, and deploy cutting-edge machine learning approaches to integrate large-scale multi-modal phenotypic datasets.
  • Develop workflows to enable post-GWAS analysis of results, e.g., fine-mapping, Translational genetics deep dives for annotation and exploration of candidate genes.
  • Design statistical methods to improve rare variant burden tests and power in longitudinal phenotypes.
  • Develop ML models for imputing disease-relevant phenotypes from high-content clinical imaging datasets (MRI, PET-CT).
  • Develop ML methods for disentangling and genetically interpreting axes of variation in complex phenotypes.
  • Use large language models to extract disease-relevant information from medical records.
  • Explore generative models of small molecules, biologics, and/or oligonucleotide therapeutics in data modalities such as 2D and 3D representations for hit-to-lead drug discovery.
  • Develop new geometric deep learning methods to better characterize nuanced molecular properties and relationships.
  • Identify and prototype novel microscopy-driven phenotyping workflows, including hardware acquisition, post-processing, and featurization.
  • Develop robust software tooling to support deployment of new and existing methods for general use by Insitro scientists.
  • Optimize existing microscopy acquisition methods in both hardware and software, using ML feature outputs to benchmark improvements.
Desired Qualifications
  • First-hand experience with biological data, preferably using computational approaches.
  • Passion for learning how to work with diverse functional genomic assays (RNA/DNase/ATAC/ChIP-seq, etc).
  • Interest in learning how to analyze single-cell RNA-seq data.
  • Solid understanding of computational chemistry, including virtual screening (classic QSAR modeling, structure based drug-discovery), library design, etc.
  • Demonstrated ability to use and develop cutting edge statistical and machine learning methods inspired by real problems.
  • Experience with machine and deep Learning frameworks (e.g., scikit-learn, PyTorch, etc.).
  • Demonstrated ability to write high-quality, production-ready code (readable, well-tested, with well-designed APIs).
  • Experience in Linux environment, database languages (e.g., SQL, No-SQL) and version control tools such as Git.
  • Publications of high-quality work in relevant computational biology, bioinformatics, systems biology, life sciences, or biomedical venues.
  • Familiarity with the SciPy/PyData ecosystem (numpy, pandas, scipy, dask etc.).
  • Familiarity with cloud computing services (AWS or GCP).
  • Familiarity with statistical analysis software, e.g., R.

Insitro uses machine learning and biology to speed up drug discovery and development by building predictive models from large-scale data to forecast which research paths will yield medicines, reducing costly failures in pharmaceutical R&D. It integrates data generation with ML to guide decisions on targets, molecules, and experiments, helping researchers focus on the most promising options. The company blends biology and ML in integrated workflows, supported by scientists and engineers who generate and interpret data for drug candidates. Its goal is to make pharmaceutical R&D more efficient and effective, shortening timelines and increasing the chances of delivering useful medicines, often via predictive insights or co-development partnerships with pharma companies.

Company Size

201-500

Company Stage

Series C

Total Funding

$643M

Headquarters

San Francisco, California

Founded

2018

Get referred to Insitro

See people who can refer or advise you

Simplify Jobs

Simplify's Take

What believers are saying

  • June 8, 2026 ADA data advanced CTRO-1013 toward first-in-human trials this year.
  • BMS added ALS-2 and ALS-3 on March 23, 2026, bringing $10 million milestone cash.
  • January 12, 2026 acquisition of CombinAbleAI strengthened insitro’s biologics design and developability stack.

What critics are saying

  • May 7, 2025 layoffs cut 22% of staff, exposing a fragile biotech funding model.
  • CTRO-1013 remains preclinical; a failed IND or Phase 1 would damage the core thesis.
  • Without CTRO-1013 clinical proof, insitro becomes a platform company with no product engine.

What makes Insitro unique

  • Daphne Koller’s Virtual Human turns human genetics into target selection faster than intuition.
  • TherML, launched January 12, 2026, spans small molecules, oligonucleotides, antibodies, and biologics.
  • BMS kept expanding ALS work since 2020, validating insitro’s platform with repeat target nominations.

Help us improve and share your feedback! Did you find this helpful?

Benefits

Excellent medical, dental, and vision coverage

Excellent mental health and well-being support

Open vacation policy

Access to free onsite baristas & cafe with daily lunch and breakfast

Access to free onsite fitness center

Commuter benefits

Paid parental leave

Competitive pay and 401(k) matching

Flexible work schedule (on site and remote)

Growth & Insights and Company News

Headcount

6 month growth

-1%

1 year growth

-3%

2 year growth

0%
MedicaEx
Jun 9th, 2026
insitro presents new data demonstrating its ai-discovered MASH candidate shows anti-fibrotic signal beyond liver-fat reduction at the American Diabetes Association 86th Scientific Sessions.

insitro presents new data demonstrating its ai-discovered MASH candidate shows anti-fibrotic signal beyond liver-fat reduction at the American Diabetes Association 86th Scientific Sessions. * Findings from insitro's lead therapeutic CTRO-1013 program demonstrate that liver-targeted silencing of IRS1 lowered circulatory biomarkers associated with fibrosis progression and hepatocyte injury in an industry-standard fibrogenic preclinical mouse model of MASH * Data presented today at ADA provide converging evidence from human genetics and preclinical models supporting IRS1's effect on liver fat and liver fibrosis * The findings extend insitro's AI-driven discovery of IRS1 - the strongest of 480 genetic signals for liver fat - and support the development of CTRO-1013 toward first-in-human trials this year NEW ORLEANS-(BUSINESS WIRE)-insitro, the physical AI company unlocking causal human biology, today presented new preclinical data from its metabolic dysfunction-associated steatohepatitis (MASH) therapeutic development program during a podium presentation at the American Diabetes Association (ADA) 86th Scientific Sessions. The data demonstrate that liver-targeted silencing of IRS1 reduced fibrosis progression and injury-associated circulatory biomarkers in an industry-standard fibrogenic preclinical mouse model, with effects that appear partly independent of reductions in liver fat. The findings support the continued advancement of CTRO-1013, the company's liver-targeted IRS1 candidate, through IND-enabling studies toward first-in-human trials this year. The presentation, "Hepatic IRS1 as a Therapeutic Target in MASLD: AI/ML Driven Genetic Discovery and In Vivo Therapeutic Validation," was delivered by Santhosh Satapati, Ph.D., Senior Director of Metabolic Diseases at insitro. Metabolic dysfunction-associated steatotic liver disease (MASLD) is a common chronic liver disease marked by excess fat accumulation in the liver. MASH is the progressive form of MASLD in which fat accumulation is accompanied by inflammation, liver injury, and fibrosis. Fibrosis - the scarring that can progress to cirrhosis, liver failure, and liver cancer - is one of the strongest predictors of poor outcomes for patients with MASH. Yet finding high-confidence targets has been difficult because the most relevant liver phenotypes, fibrosis among them, are hard to measure at the scale needed for human genetic discovery. Human genetic analyses revealed an IRS1-associated fibrosis signal that persists even after adjusting for liver fat, suggesting a fibrotic mechanism independent of steatosis. To evaluate this genetic hypothesis, insitro tested liver-targeted IRS1 silencing in preclinical models. Data presented today showed significant reductions in TIMP-1 and CK-18 levels, key biomarkers associated with fibrosis progression and hepatocyte injury. IRS1 inhibition also reduced liver fat in AMLN-DIO mice. Together, these results reinforce prior insitro data demonstrating improvements in liver-injury endpoints and fibrosis-related gene expression, establishing a consistent role for IRS1 across both human genetics and preclinical models. "MASH remains an area of significant unmet need, with a dearth of interventions capable of fundamentally shifting the disease trajectory," said Daphne Koller, Ph.D., founder and CEO of insitro. "The primary obstacle has been bridging the gap between early metabolic indicators, like de novo lipogenesis, and the fibrotic processes that dictate clinical outcomes. Utilizing our Virtual Human(TM) platform, we applied AI-derived liver fat and fibrosis phenotypes from multimodal human data to identify causal targets, such as IRS1, where metabolic stress and fibrosis biology appear to converge." To translate this target discovery into clinical development, insitro is advancing CTRO-1013, an investigational liver-targeted siRNA therapeutic designed to selectively silence IRS1. "IRS1 stands out because it bridges what have often been treated as two separate problems: metabolic dysfunction and fibrotic scarring," said David Lloyd, Ph.D., Senior Vice President of Metabolic Disease and Translational Pharmacology at insitro. "By targeting IRS1 with CTRO-1013, we see improvements in liver fat alongside encouraging shifts in biomarkers linked to fibrosis progression. While still preclinical, these data suggest that IRS1 inhibition could offer a path to meaningfully altering the trajectory toward severe, end-stage liver disease." insitro identified IRS1 through analysis of human phenotypic, biomarker, and genetic data, including UK Biobank imaging and clinical datasets linked to liver fat and adipose biology. Genetic analyses prioritized IRS1 as a therapeutic target associated with liver fat biology and revealed a fibrosis-associated signal that persisted after adjustment for fat. The company then validated the target experimentally. A potent, selective, liver-targeted siRNA against IRS1 suppressed lipid accumulation in human hepatocytes. In diet-induced obese mice, a single dose reduced hepatic IRS1 expression by 85%, de novo lipogenesis by 59%, and hepatic triglycerides by 45% after one month, while significantly improving NAS score versus vehicle. Liver-targeted IRS1 silencing also led to significant reductions in TIMP-1 (37%) and CK-18 (68%) after 16 weeks of treatment (1 mg/kg, Q4W) in a standard fibrogenic preclinical mouse model compared to vehicle-treated controls. In cynomolgus monkeys, liver-targeted IRS1 silencing produced sustained target reduction over 90 days without affecting glucose homeostasis. "The strength of these findings lies in the consistency of the signal - from human cells to preclinical models and non-human primates - across lipogenesis, liver injury, and fibrosis," said Philip Tagari, Chief Scientific Officer of insitro. "This breadth of evidence is critical; it indicates that IRS1 modulation impacts disease pathology through pathways that are distinct from simple fat clearance, reinforcing our therapeutic rationale as we advance CTRO-1013 toward the clinic." CTRO-1013 was designed from the outset for tissue-selective delivery to the liver, concentrating activity where it is needed while limiting systemic exposure. According to Satapati, the integration of AI-derived phenotypes, human genetic data, clinical imaging, and rigorous preclinical testing provides a clearer, more human-relevant view of IRS1 inhibition than traditional animal models alone. "The anti-fibrotic signals observed in the data appear to function through mechanisms that are partially independent of liver-fat reduction, providing a vital insight the team is continuing to investigate." Additional mechanistic and histological analyses are ongoing. insitro is continuing IND-enabling studies for CTRO-1013 and expects the program to enter first-in-human clinical trials this year. Presentation Details Title: Hepatic IRS1 As a Therapeutic Target in MASLD: AI/ML Driven Genetic Discovery and In Vivo Therapeutic Validation Abstract Number: 1331-OR Presenter: Santhosh Satapati, PhD Meeting: American Diabetes Association 86th Scientific Sessions Location: New Orleans, Louisiana About insitro insitro is the physical AI company unlocking causal human biology, founded and led by AI pioneer Daphne Koller. By generating the world's largest integrated multimodal corpus of human and cellular data, Medicaex has built the Virtual Human(TM)- a genetically anchored causal AI engine that reveals how disease begins, progresses, and can be resolved. Its platform enables Medicaex to precisely identify causal genetic drivers and deploy its TherML AI platform to design optimal medicines, advancing a broad pipeline of therapeutics for neuroscience and metabolic diseases. This industrialized architecture creates a self-learning loop: with every biology Medicaex onboard, its predictive models grow smarter, accelerating discovery across scales of biology. Backed by more than $750M in capital from world-class investors like a16z, ARCH, BlackRock, Casdin, CPP, Foresite, GV, SoftBank, Temasek, Third Rock, T. Rowe Price - including ~$140M in revenue from collaborations with BMS, Lilly, and Gilead - insitro is rebuilding drug discovery from an unpredictable journey into an industrialized, repeatable process with scalable impact for patients and the world.

Bionews, Inc.
Mar 30th, 2026
Developers expand collaboration to test 2 novel targets for ALS treatments.

Developers expand collaboration to test 2 novel targets for ALS treatments. Insitro, Bristol Myers Squibb first struck deal in 2020 to use AI in lab research * Insitro and Bristol Myers Squibb are expanding a research collaboration deal, in place since 2020, to find new ALS treatments. * The two companies are using AI and patient-derived cells to identify biological targets for ALS therapies. * Two new targets aim to correct TDP-43 protein abnormalities, improving nerve cell health. Insitro and Bristol Myers Squibb have expanded their ongoing collaboration to develop new treatments for amyotrophic lateral sclerosis (ALS), adding two newly identified disease targets to their ongoing research effort. The two drug companies have been working together since 2020 to uncover biological changes driving ALS that may be reversed with therapies, using a combination of artificial intelligence (AI) and lab-grown human cells. The new targets, ALS-2 and ALS-3, build on previous work in the collaboration, which has already advanced a first target, dubbed ALS-1, into early development. The companies are already exploring different classes of therapies to act on the same disease target. "By expanding our collaboration with Bristol Myers Squibb, we are broadening our approach to tackling this devastating disease, with a set of compelling targets that address its fundamental mechanisms, with the goal of delivering disease-modifying therapies to the many patients who cannot wait," Daphne Koller, PhD, Insitro's founder and CEO, said in a company press release. A $10 million payment was tagged to target selection, according to Insitro. Recommended Reading ALS is caused by the progressive loss of motor neurons, the nerve cells that control voluntary movement. While the exact mechanisms leading to motor neuron degeneration and death are not fully understood, abnormal clumps of the TDP-43 protein are believed to play a role. Under normal conditions, TDP-43 is found in the nucleus, where genetic material is stored, and helps cells correctly process genetic instructions needed to make proteins. In about 97% of all ALS cases, however, TDP-43 forms clumps outside the nucleus and is no longer able to perform this function properly. AI helps create 'data-driven map' showing impact of ALS treatments. In the initial phase of the duo's collaboration, Insitro generated motor neurons from induced pluripotent stem cells (iPSCs), which are adult cells reprogrammed into a stem cell-like state. By using cells derived from patients or carrying mutations implicated in ALS, researchers were able to study the disease in a controlled setting and generate large datasets for analysis. Using its AI-based platform, the company analyzed these data alongside large clinical datasets to identify patterns and pinpoint biological pathways that are consistently disrupted in ALS and could be targeted with new therapies. "We are driven by a sense of urgency to translate our biological insights into meaningful clinical outcomes for the ALS community," said Koller. "Our platform allows us to build a data-driven map of the impact of ALS on motor neurons and identify novel drivers of neurodegeneration." So far, three new targets have been identified that may modulate the effects of TDP-43 abnormalities. These could have the potential to treat the vast majority of ALS patients, according to the company. Validation experiments using patient-derived motor neurons showed that modulating these targets increased neurite growth, a key indication of nerve cell health. This was accompanied by a reduction of the genetic abnormalities seen when TDP-43 is missing in the nucleus, and an increase in normal protein production. Under the terms of the agreement, Insitro has received the $10 million in milestone payments for the selection of the two new targets.

Yahoo Finance
Mar 23rd, 2026
insitro receives $10M as BMS nominates two new ALS targets from AI platform

insitro has expanded its strategic collaboration with Bristol Myers Squibb to advance therapeutic programmes for amyotrophic lateral sclerosis (ALS). BMS has nominated two additional targets, ALS-2 and ALS-3, identified through insitro's AI-driven Virtual Human platform, joining the initial ALS-1 target nominated in December 2024. The companies will employ multiple therapeutic modalities to address the targets. insitro will advance an oligonucleotide programme for ALS-1 whilst simultaneously developing a small molecule programme for BMS. The strategy aims to maximise opportunities to impact patients quickly and effectively. insitro received a $10 million milestone payment for the selection of the two additional targets. The collaboration focuses on identifying key biological drivers to deliver disease-modifying interventions, specifically targeting processes that modulate TDP-43 mislocalization, a central disease mechanism in nearly 97% of ALS patients.

insitro
Mar 23rd, 2026
insitro and Bristol Myers Squibb collaboration expanded with nomination of new targets.

insitro and Bristol Myers Squibb collaboration expanded with nomination of new targets. Collaboration expands to include two additional therapeutic targets for the treatment of ALS discovered via insitro's AI-driven Virtual Human(TM) platform Joint effort focuses on identifying key biological drivers to deliver disease-modifying interventions for ALS patients SOUTH SAN FRANCISCO, Calif. - insitro, the AI therapeutics company built on causal biology, today announced the expansion of its strategic collaboration with Bristol Myers Squibb (NYSE: BMY) to advance a broadened portfolio of therapeutic programs for amyotrophic lateral sclerosis (ALS). The collaboration is focused on accelerating and delivering disease-modifying interventions designed to address the underlying biological drivers of ALS. BMS has nominated two additional targets, ALS-2 and ALS-3, which were identified through insitro's Virtual Human(TM) platform. These join the initial target, ALS-1, nominated by Bristol Myers Squibb in December 2024. The companies will leverage multiple therapeutic modalities to address the nominated targets. insitro will advance its own oligonucleotide program for ALS-1 while simultaneously progressing a small molecule program for BMS for ALS-1. This multimodality development strategy is designed to maximize the opportunities to impact patients as quickly and effectively as possible. insitro received a $10 million milestone payment in connection with the selection of the two additional targets. "We are driven by a sense of urgency to translate our biological insights into meaningful clinical outcomes for the ALS community," said Daphne Koller, Ph.D., founder and CEO of insitro. "Our platform allows us to build a data-driven map of the impact of ALS on motor neurons and identify novel drivers of neurodegeneration. By expanding our collaboration with Bristol Myers Squibb, we are broadening our approach to tackling this devastating disease, with a set of compelling targets that address its fundamental mechanisms, with the goal of delivering disease-modifying therapies to the many patients who cannot wait." Leveraging insitro's Virtual Human(TM) causal biology discovery platform, the company has identified a series of high-impact targets that play a central role in the biological mechanisms underlying ALS. By integrating massive-scale, human-derived cell data with machine learning, the Virtual Human(TM) allows for the mapping of disease drivers with unprecedented resolution, specifically focusing on processes that modulate the effects of TDP-43 mislocalization - a central disease mechanism in nearly 97% of ALS patients. In validation experiments using iPSC-derived motor neurons, modulation of these targets rescues neurite growth in cellular models of ALS - a significant milestone that reflects structural repair in human neurons. This is accompanied by reduction of the cryptic exons and restoration of the corresponding full-length transcript by a significant amount, reversing key markers of disease pathology that occur broadly in ALS patients. This also provides strong evidence supporting the potential of these targets to lead to disease-modifying therapies. About insitro insitro is the physical AI company unlocking causal human biology, founded and led by AI pioneer Daphne Koller. By generating the world's largest integrated multimodal corpus of human and cellular data, Insitro, Inc. has built the Virtual Human(TM)- a genetically anchored causal AI engine that reveals how disease begins, progresses, and can be resolved. Its platform enables Insitro, Inc. to precisely identify causal genetic drivers and deploy its TherML AI platform to design optimal medicines, advancing a broad pipeline of therapeutics for neuroscience and metabolic diseases. This industrialized architecture creates a self-learning loop: with every biology Insitro, Inc. onboard, its predictive models grow smarter, accelerating discovery across scales of biology. Backed by ~$800M in capital from world-class investors like a16z, ARCH, Blackrock, Casdin, CPP, Foresite, GV, Softbank, Temasek, Third Rock, T. Rowe Price - including ~$150M in revenue from collaborations with BMS, Lilly, and Gilead - insitro is rebuilding drug discovery from an unpredictable journey into an industrialized, repeatable process with scalable impact for patients and the world. Media Contact Eric McKeeby [email protected]

Business Wire
Feb 26th, 2026
insitro appoints Joe Hand as chief people officer to advance talent strategy

insitro, an AI therapeutics company, has appointed Joe Hand as Chief People Officer to lead its global talent strategy as it scales operations and advances drug discovery programmes toward clinical trials. Hand brings over two decades of life sciences leadership experience, including nearly a decade at Celgene, where he served as Chief Human Resources Officer. At Celgene, he led HR operations during major transactions, including the $74 billion Bristol Myers Squibb acquisition and the $9 billion Juno acquisition. He also served as Chief Administrative Officer at Phathom Pharmaceuticals, building its HR infrastructure from inception. Founded by AI pioneer Daphne Koller, insitro uses integrated AI and biology platforms to discover medicines for neuroscience and metabolic diseases. The company has raised approximately $800 million from investors including a16z, ARCH and Softbank.

INACTIVE