Technical Lead, Scientific Computing Platform (Python) - Part-time - Worldwide

Jane Cervantes, Recruiter

• Required expertise: Python, FastAPI, Pydantic, PostgreSQL, Docker, automated testing and continuous integration. The candidate should have delivered scientific or engineering software with reliable APIs, workflow orchestration and traceable data.
• Useful experience: Temporal, Kubernetes, GPU/HPC workloads, PyTorch or JAX, Julia and integration with C++/Fortran solvers. Familiarity with AI coding assistants and their verification is valuable.
• Initially 600 hours over 20 weeks, including 150 hours for the Research Workbench.

About the role

TVT seeks a hands-on software architect to lead the development of a fusion energy and materials science system from architecture through tested release. The candidate should bring strong experience in Python, API design, relational data, containerization, testing and software delivery. Experience with FastAPI or a comparable framework and with typed data contracts is important.

The candidate will establish the Research Workbench, supervise six specialist rTOs and own software delivery and property integration. The candidate will assign work, review interfaces and resolve dependencies across the team. For physical and materials property data, the candidate will ensure that workflows record the source, version, units, uncertainty and conditions under which each value applies. Domain specialists will assess whether the data and models are suitable for the intended engineering use.

The candidate will review the platform’s primitives, artifact definitions and archetypes with the specialist rTOs. Check that the basic building blocks have clear meanings and types, that records and datasets carry enough information to reproduce and review the work and that reusable workflow patterns state their inputs, outputs and acceptance criteria. Prioritize changes where the definitions create ambiguity or prevent reuse across workflows.

The candidate will define how the platform maps its concepts to external ontologies, data schemas and solver vocabularies, and set the conventions for identifiers, versioning and validation. The candidate will review mappings used across multiple workflows and make sure gaps or ambiguities are visible. Specialist rTOs will develop domain mappings and work with scientific experts to validate their meaning.

The candidate will oversee physics solver integration with domain specialists. For example, The candidate may lead the design and testing of a workflow that passes a neutron spectrum from OpenMC to FISPACT-II for an activation inventory calculation. The workflow must check energy group boundaries, units and normalization, preserve the exact inputs and recover from an interrupted run.

The candidate will coordinate the DAR (Domain Architecture Reviewers) review schedule, prepare evidence for review and track agreed actions through closure. The DARs retain authority over their independent findings and assessment conclusions. The CAE Expert retains engineering coordination.

The candidate’s deliverables will include architecture decisions, reusable data contracts, documented mappings, an integration backlog, tested release procedures and operating documentation. The platform must enforce permissions, preserve a traceable record of calculations and distinguish capabilities accepted for use from research candidates still under evaluation.

Experience with scientific computing, ontology or schema mapping, Temporal, cloud CPU or GPU workloads, Nebius or machine learning systems is useful. Bounded program synthesis and Monte Carlo tree search are additional advantages. Please share examples of systems The candidate delivered and explain The candidate’s contribution, the technical trade-offs and how the candidate worked with scientific users. Include an example of handling invalid data, an external failure or a change in requirements. ⓘ View potential timezone restrictions

Selection evidence. A paid exercise will cover a reproducible calculation, invalid input rejection and recovery after an interruption. We will assess maintainability, communication and judgement in designing scientific interfaces. Please provide the candidate’s availability, time zone, relevant work and rates.

About the company

TVT is a deep-tech industrial technology venture transforming global energy generation and heavy process industries. Grounded in advanced materials science and physics-informed computational engineering, TVT’s core mission is to achieve among the world’s lowest Levelized Cost of Electricity (LCOE) alongside ultra-low-cost industrial process heat through transformative, intrinsically safe fusion energy systems engineered for flexible baseload operations.

Beyond power generation, TVT engineers integrated industrial-scale infrastructure-including high-temperature energy storage, thermal recovery networks, and process-coupling interfaces-designed to maximize overall thermodynamic efficiency and decarbonize hard-to-abate industrial manufacturing.

Core Product Divisions

• TVT-Power: Complete commercial fusion energy plants engineered around the TVT-500, a proprietary Tokamak-Stellarator hybrid reactor concept. The baseline plant design targets delivering 500 MWe of net electricity from 1,250 MWth of thermal energy (~40% net electric efficiency) with continuous, dependable baseload availability.
• TVT-Storage: Long-duration thermochemical energy storage systems engineered for continuous energy dispatch and high-temperature thermal management without dependence on specific geographic or geological formations.
• TVT-Heat: High-temperature industrial heat recovery, cascaded thermal loops, and precision process-temperature matching (delivering industrial heat up to 500°C) with comprehensive cooling and residual-heat integration.
• TVT-Process: Modular industrial balance-of-plant systems that synthesize electricity, direct thermal energy, and process technologies to substantially increase the conversion efficiency of heavy chemical and manufacturing facilities.

Python fast apiPythonPostgreSQLDocker