Skip to content
PROMAXBIOSCIENCE
Careers

Build the platform, not just the program.

We are a small team working on a hard problem in one of the most demanding areas of drug development. If you believe R&D itself can be systemised rather than merely automated, we should talk.

Working here

Small team. Long horizon. Real ownership.

One problem, deeply

We are not running twenty programmes in parallel. Depth is the advantage of a company our size, and we protect it.

Evidence over hierarchy

Data settles arguments here. Anyone who can design the experiment that distinguishes two hypotheses gets the floor.

Process is respect

A method that only works in its inventor’s hands is not a method. We document so that the next person can repeat the result.

Open positions

Five AI research roles,
one for each of the first five layers.

All open roles are for bioinformatics and computational biology backgrounds, with location unrestricted. Each role maps to one layer of our six-layer R&D system; the sixth, the governance layer, is covered by the existing team and is not open this round.

L1 · Device layerFull-time

AI Researcher, Device Layer

Location: unrestricted

Turn laboratory instruments into computable data sources, not black boxes that need manual transcription.

What you would do

  • Map the output of chromatography, mass spectrometry, plate readers, sequencers and automated workstations onto one set of acquisition fields, units and status codes.
  • Work with instrument vendors and our engineering team on parameter models for methods, batches, channels and QC flags, so the same experiment stays reproducible on different instruments.
  • Decompose protocols into executable, reversible step sequences for liquid handlers and robotic arms.
  • Write data-consistency checks and acceptance plans for every newly connected instrument, leaving an auditable integration record.
  • Convert device-side anomalies — drift, failed batches, consumable lot variation — into structured signals the layers above can use directly.
L2 · Driver layerFull-time

AI Researcher, Driver Layer

Location: unrestricted

Absorb vendor and firmware differences so every instrument speaks one language inside the system.

What you would do

  • Maintain a cross-vendor, cross-firmware driver adaptation layer that exposes a single device and acquisition interface to the layers above.
  • Own driver version control and regression testing, automatically diffing acquisition output after firmware upgrades to catch silent changes.
  • Design orchestration, checkpointing and retry for long acquisitions such as sequencing, imaging and continuous culture.
  • Define field mappings and unit conventions together with the device and data layers, eliminating competing definitions of the same measurement.
  • Document integration and troubleshooting so a new instrument goes live in days rather than weeks.
L3 · Data layerFull-time

AI Researcher, Data Layer

Location: unrestricted

Make every experiment retrievable, reproducible and modellable without depending on anyone’s memory.

What you would do

  • Design metadata and master-data models — project, target, molecule, batch, sample, assay — and give the company one data vocabulary.
  • Build preprocessing and QC pipelines that detect missing values, batch effects and outlier samples while preserving a complete processing trail.
  • Establish dataset versioning and annotation so every training set can be traced back, layer by layer, to raw signal.
  • Define completeness, consistency and timeliness metrics for each data domain, and track them on a monitoring dashboard.
  • Work with IT and security on tiered access, backup policy and restore drills.
L4 · Application layerFull-time

AI Researcher, Application Layer

Location: unrestricted

Turn R&D process into system process, so conclusions stay with the organisation rather than with one person.

What you would do

  • Model initiation, protocol, execution, analysis and reporting as configurable workflows that different programs can reuse.
  • Build analysis and visualisation tools in which antibody screening, LNP formulation and potency data are comparable in a single view.
  • Develop automated reporting and knowledge management that turn program conclusions into searchable organisational memory.
  • Design alerting and diagnostic rules — potency loss, encapsulation outliers, batch drift — and feed every trigger into the AI-layer decision chain.
  • Run user research with the lab teams, converting methods that only work in their inventor’s hands into handover-ready system flows.
L5 · AI layerFull-time

AI Researcher, AI Layer

Location: unrestricted

Decide what the next batch of experiments should be, under a finite experimental budget.

What you would do

  • Build and validate models for sequence design, lipid component screening and condition optimisation, stating each model’s validity range and failure conditions.
  • Design experiment-decision logic, back-testing recommendations against historical data and quantifying the gain in success rate and cycle time.
  • Establish model validation — data splits, external validation, drift monitoring — so algorithm validation and system validation leave the same audit trail.
  • Bring people, pipeline, material and budget data into one decision view that supports program prioritisation and resourcing.
  • Define falsifiable hypotheses together with the wet-lab teams, so model output never stops at correlation.

Common requirements

All five roles share the same professional baseline; the layer decides where the work sits.

  • M.Sc. or above in bioinformatics, computational biology or a closely related field; Ph.D. preferred.
  • Fluent in Python or R, with hands-on experience handling multi-omics, high-throughput screening or process data.
  • Experience taking an analysis pipeline or model into production and working alongside wet-lab scientists over time.
  • Able to read experimental protocols and assays, and to translate a biological question into a computable definition.
Apply by email

hr@promaxbioscience.cn

No matching role?

Tell us what you would build.

Send a short note describing the problem you want to work on, and a CV.

Email our team