Technical Project Manager

Nuclir is hiring a Technical Project Manager to own delivery across the systems that keep the intelligence layer current: ingestion of public and licensed sources, signal resolution, model training and evaluation, and the regional path that serves the API and the Console. You hold the sequence, the dependencies, and the definition of done for work that crosses research, data engineering, machine learning, and infrastructure. This is a technical role rather than a coordination role. You read the design document, ask the question that exposes the unstated dependency, and write the plan the team then works to. Much of the value is in sequencing: a source integration that unblocks a model, an evaluation that has to run before a threshold ships, a delivery change that cannot land in the middle of a customer trial, a vendor dependency with a long lead time that has to be started two quarters early. Expect ambiguity. Nuclir is early, the surface is wide, and the constraints are physical: sources publish on their own cadence, point-in-time correctness is not negotiable, and freshness is the product rather than a feature of it. You decide what ships this month and what is honestly deferred, and you are the person who says plainly which is which. What you'll do Own two or three concurrent workstreams end to end: scope, sequence, dependencies, risk, and a date the team can defend to a customer. Turn research and engineering intent into a written plan - milestones, owners, entry and exit criteria, and the evidence that closes each one. Run planning, status, and retrospectives as short written artifacts that a reader understands without attending a meeting. Track dependencies between source ingestion, entity resolution, model evaluation, and delivery, and surface a slip the week it appears rather than the week it lands. Hold the release path with infrastructure: what ships, behind which flag, with which rollback, and which latency and freshness measurements are read afterwards. Maintain the delivery risk register, including single-owner work, sole-source vendor dependencies, and the licensing or export control questions that gate a source. Represent schedule reality to leadership and to customers in a trial, including when the honest answer is later than the one being asked for. What we're looking for Five or more years managing technical programs in software, data infrastructure, or a comparable engineering organization. Fluency with the systems you coordinate: distributed data pipelines, model training and evaluation loops, and APIs with a regional delivery path. You do not need to write production code, but you do need to read a design and find the gap in it. A record of shipping work that crossed at least three teams, where the sequencing and dependency calls were yours to make. Written communication as the default. Plans, decisions, and status land as documents, not as verbal updates that only the attendees hold. Comfort planning against instrumentation - latency, freshness, correctness, recovery - rather than against percentage complete. Judgment on scope: you can say what is out, defend it with the constraint behind it, and reopen it when the constraint changes. Fluent English and overlap with US working hours, with the provenance discipline to record why a decision was made and when. #J-18808-Ljbffr Nuclir Data