Position: Senior Staff Engineer, Embedded AI
Team: Embedded AI Engineering
Level and grade: L6, Associate Director equivalent on the Embedded AI Engineering band
Position type: Full time, permanent
Location: India, with regular time on client sites
Reports to: Chief Executive Officer
Travel: 25 to 30 percent, mostly to client sites in India
About Us
APLYD helps governments, multilateral institutions, development finance institutions and foundations use AI in public systems. Most public-sector AI stops at the pilot. Our work is getting it into everyday service delivery and keeping it running once we leave.
Athena Infonomics has done this work for years. In 2026 we set it up as its own company. We cover strategy and readiness, field data and last-mile reach, design and build, evaluation and audit, and scale and production.
440+ engagements · 240+ global clients · 85+ specialists · 7 countries · 5 continents
The Role
You are the senior engineer on APLYD's work. You design and build the systems public institutions run on, and the models that sit inside them, for clients whose data is incomplete and whose infrastructure you do not control. You set the technical standard for everyone who joins after you.
Most of what we build is not a model. Registries, credential and wallet layers, group financial records, synchronisation for intermittent connectivity, offline-first clients and the migrations that keep them alive are the bulk of the work, and they have to be engineered properly before any model on top of them is worth anything.
Embedded describes the way of working, close to the institution and its decisions. It does not require sitting in the institution's office or country.
We do not staff a specialist for each component. Everyone here writes production software and everyone here builds models; the levels are separated by the scope a person carries rather than by the technology they work on.
This is a hands-on role. You write code every week and you train models every month, and you continue to do both as the team grows.
Core Job Responsibilities
- Design and build production systems. Service and API design, relational data modelling, schema migrations that run against live government databases, synchronisation for clients that are offline for days, and the identity and access layer underneath all of it.
- Build models where the problem needs one. Frame it, assemble and label the data, choose the approach, train, evaluate and deploy. Classical machine learning, language models, retrieval or forecasting, whichever the problem actually needs.
- Own evaluation before you own the model. Construct the test sets, choose the metrics, find the leakage, and report performance by subgroup rather than as a single headline number.
- Decide what is and is not a machine learning problem, and say so to the client before anyone has spent money on it.
- Own the data layer: ingestion, validation, reconciliation against source systems, pipelines and feature engineering on data that arrives late and inconsistent.
- Own production. Reliability, monitoring, alerting, incident response, model drift and retraining, and the rollback path for when something starts behaving badly on a Friday evening.
- Hold the engineering discipline: tests that mean something, code review, continuous integration, release management, backward compatibility, and documentation a stranger can follow.
- Own security and privacy in systems holding personal data at population scale. Authentication and authorisation, secrets handling, encryption, audit trails, and data minimisation designed in rather than retrofitted.
- Make the compute and infrastructure decisions: scaling, latency, cost and reliability, designed for constrained hosting, limited connectivity, procurement rules that limit what you can use, and long periods without active monitoring.
- Work closely with the institution. You spend real time in the departments we serve, with the people who will use what you build and the people who will run it after we leave. Much of what makes public-sector systems fail is only visible in the office: undocumented workarounds, fields that are always left blank, approvals nobody mentions until you need them.
- Prototype quickly when the situation calls for it, using coding agents and modern tooling, and be clear which artefacts are temporary and which will be hardened.
- Support business development. You scope on incomplete information, give a number and a timeline you can stand behind, and sometimes build the prototype that makes a proposal concrete.
- Package for handover. Systems, models and documentation that a delivery team or a government IT unit can run, extend and retrain without us.
- Set the technical standard and mentor the engineers who join after you.
Qualifications and Competencies
- Ten or more years in software engineering, including production machine learning. You can describe systems you designed and models you trained, on what data, evaluated how, and what happened once they were live.
- Deep conventional engineering. You have designed and run services other systems depend on: API design and versioning, relational data modelling, transactions, schema migrations, caching and queues.
- You have trained, evaluated and deployed models yourself, recently. Classical machine learning as the base, plus real depth in at least one of language models and retrieval, forecasting, or computer vision.
- Python at production standard, with scikit-learn and at least one of PyTorch or TensorFlow used on work that shipped. Fluency in a second language used for backend work is expected.
- You design evaluation before you build: test set construction, metric choice, leakage, class imbalance, and performance across subgroups.
- SQL to a serious standard, and the data engineering needed to make government data usable, including reconciliation against the source system.
- Engineering hygiene as a matter of habit: testing, code review, continuous integration, release discipline, and debugging production systems you did not write.
- Security and privacy engineering for systems holding personal data: authentication, authorisation, secrets, encryption and audit.
- Cloud platforms, containers, GPUs, scaling and cost control, and deployment into environments you do not fully control.
- You use AI coding tools and agents fluently and know where they stop being reliable.
- You understand the commercial side of delivery: what a project was sold for, what it cost to deliver, and how technical choices affected that.
Also useful, though we will not screen on it
- Systems built for government, public-sector or large institutional environments.
- Digital public infrastructure: registries, identity, credentials, wallets or data exchange layers.
- Offline-first or low-connectivity systems at scale.
- Fine-tuning or serving open-weight language models on constrained infrastructure.
- Multilingual, low-resource or last-mile data.
- Mentoring engineers or leading a small technical team.
Additional Requirements
- This position requires successful completion of a reference check and employment verification.
- The successful candidate must not be subject to employment restrictions from a former employer, such as a non-compete, that would prevent performance of the responsibilities described.
- Candidates must declare any current or recent engagement with a government, multilateral or development finance institution that could present a conflict of interest.
APLYD’s Work Culture
At APLYD, we function in an outcomes-based work environment with flexible hours and a high level of autonomy. Professional development and thought leadership are key elements of our business model: we support our team members’ professional growth through on-the-job training, and we encourage the cultivation of our colleagues’ personal brands through participation in panels, events, publications, and other thought-leadership opportunities. We embrace a transparent, open work environment with meaningful leadership pathways for those with inventive ideas and initiatives.
APLYD is an Equal Opportunities Employer
APLYD, part of the Athena Infonomics group, is an equal opportunity employer with a commitment to diversity. All qualified applicants will receive consideration for employment without regard to race, color, religion, sex, pregnancy, sexual orientation, gender identity, national origin, age, protected veteran status, or disability status.
AI Proficiency and Responsible Use
Proficiency in the responsible and sophisticated use of AI is a mandatory requirement for all roles, across all levels and functions at APLYD and Athena Infonomics.
