Avanc is an early-stage deep-tech company based in Bristol, UK. Co-founded by a biologist, Dr Eleanor Jaskowska and an engineer, Lizzie Melhuish, we are a multidisciplinary team building novel optical sensing instrumentation that makes biological risk in the environment visible in real time. Avanc is currently funded by grants and investment from Zinc VC. We are about to open our pre-seed fundraising round. This is an exciting time to join us as one of two first hires.
Machine Learning Engineer
Full or Part Time | Permanent | Bristol
Start December 2026/January 2027
We are hiring our first Engineer to work on the machine learning model at the core of our biosensing platform.
We are hiring an experienced Machine Learning Engineer to built, train, optimize and deploy our models reliably. Some experience and ability in Cloud Engineering and Software Engineering is desirable to architect and build our software stack from the ground up. This is a role, with real ownership. You will own the entire digital pipeline that turns raw data from our physical units into actionable, real-time alerts.
You'll be reporting to the CTO/Lead Engineer and working alongside a R&D Scientist building the sensor hardware. This is a fantastic opportunity for a commercially curious and self-motivated engineer who also wants to develop their knowledge of how an early stage startup works.
The Role
Working closely with our CTO and R&D Scientist you will:
own the classification system at the heart of our platform
convert raw spectral processing to actionable user alerts
work with small, noisy datasets that are tightly coupled to the underlying physical instrumentation.
The Scope
Spectral signal processing
Baseline correction, cosmic ray and noise filtering, normalisation, peak extraction and feature engineering on 1D spectra.
Matrix-specific background models.
Classification and validation
Build and train a classifier over a reference library containing both bacterial and chemical targets.
Design validation. Classes are imbalanced and accuracy alone is not a sufficient metric.
Define operational targets.
Report per-class, with confusion matrices and ROC curves.
Handle systematic differences in signal quality between classes.
Establish the ground-truth and collection requirements to shape the laboratory work.
Contribute to the product roadmap, including how the library scales and how models reach deployed units.
The Stack
Language: Python, end to end.
Signal processing and ML: NumPy, SciPy, scikit-learn, PyTorch.
Backend and interface: FastAPI, Flask or a comparable lightweight framework.
Cloud: Standard AWS, GCP or Azure SDKs. Nothing exotic.
You are welcome to propose alternatives with a good technical argument behind them.
Experience: 2+ years of applied machine learning engineering experience, ideally with a background in Bioinformatics, Chemometrics or where software interacts with physical sensors.
Familiarity with 1D signal or spectra data processing or a willingness to learn.
Rigour when it comes to validation.
Bonus if you’re familiar with
Deploying models behind an API in a cloud environment with latency constraints.
Working with data coming off physical instruments, and the messiness that entails.
Chemometrics, bioinformatics or microbiology exposure.
Salary range £45-60,000 FTE depending on experience and equity package.
We are open to discussion with regards to a remuneration combining equity and salary.
Reports to: CTO
Location: Bristol, access to lab facilities is required. We are open to discussion regarding hybrid and remote working but some visits to the team in Bristol will be required
Permanent contract
You must have the right to work in the UK, we are unable to sponsor visas at this time
What you can expect from the hiring process
Successful candidates invited to 3 interviews
1st interview will assess cultural fit and motivation with CEO
2nd interview will be a technical assessment with CTO
Final stage: visit our lab in Bristol for final interview and practical challenge
Opens a new window and requires Google sign in
#J-18808-Ljbffr
