Senior ML Engineer, Design Automation
PH-003 | Founding Technical Staff | Boston, MA | Onsite | $180K – $250K base + equity
About Prahelika
Prahelika is building the domestic integration layer: 3D systems in which III-V semiconductors, state-of-the-art silicon, and novel interposers are stacked into one device, each doing what it does best. We co-design the signal path, the power tier, the thermal path, and the manufacturing process, so dies from any foundry become a system that silicon alone cannot deliver. Our first 3D systems serve defense radar and low-Earth-orbit satellite communications; the same platform extends to power electronics, data-center compute, and photonics. Backed by top-tier investors and built by engineers who demonstrated every tier of the stack as measured hardware before the company existed, Prahelika owns the layer between the transistor and the machine. Ultimate electronics through integration.
The role
Whoever owns a design here sees it through fabrication, packaging, measurement, shipping, and the next revision. Designing a 3D system means solving electrical, thermal, mechanical, materials, and manufacturing constraints at the same time, and by hand that is too slow. One of our four founding axioms is that AI is a cornerstone of our engineering: it accelerates integrated design, with physically grounded safeguards and sign-offs, so that engineers spend their time thinking and envisioning rather than turning cranks.
You will build that capability. As our first ML engineer, you design and ship the agentic design-automation system our engineers use every day: agents that drive commercial EDA and multiphysics tools, surrogate models that make thermal and EM iteration fast, and the verification layer that keeps every AI-generated result tied to physics and to measurement. This is not a research role. Your users sit ten feet away, and your software is in the loop on our first tape-outs.
Required qualifications
M.S. or Ph.D. in Computer Science, Electrical Engineering, or a related field, with 4+ years building ML or agentic systems that shipped to real users, including at least one that used LLMs as a component. Equivalent practical experience may substitute for the degree requirement.
Strong software engineering: Python, modern LLM tooling (tool use, agent frameworks, evaluation harnesses), and the discipline to make a system reliable rather than a demo.
Working familiarity with physics-based engineering tools and their data: circuit simulators, EM or FEA solvers, GDSII/STEP/netlist formats, or scientific computing generally. You do not need to be an RF engineer; you need to be able to communicate with one.
Hands-on experience in at least two of the following:
Agentic systems that drive external tools or APIs with verification, retries, and human sign-off in the loop.
Surrogate or physics-informed models (neural operators, GNNs, reduced-order models) for simulation acceleration.
EDA, CAD, or scientific workflow automation, including scripting commercial tools (Cadence, Keysight, Ansys, COMSOL, Siemens, or similar).
Evaluation and guardrail design for LLM systems in domains where a wrong answer has a physical cost.
This role involves work that may be subject to U.S. export-control laws. Prahelika assesses access requirements for the specific work and, where needed, evaluates available government authorization options before granting access to controlled technology. Program assignments are subject to applicable legal requirements.
Preferred qualifications
Prior work in semiconductor design automation, packaging, or analog/RF design.
Fine-tuning or retrieval systems over technical documentation and design data.
Responsibilities
Design, build, and operate the agentic design-automation system that our RF, analog, EM, thermal, and integration engineers use on our first products.
Integrate commercial EDA and multiphysics tools into automated flows, with physically grounded checks that gate every result before it reaches a human or a tape-out.
Build surrogate models that cut thermal, EM, and mechanical iteration time from hours to seconds, and prove they hold against full-fidelity simulation and measurement.
Work embedded with the engineering team: sit in design reviews, learn the physics, and let the tool follow the problem rather than the other way around.
Establish the software practices, evaluation harnesses, and data infrastructure that become the foundation of Prahelika's design kit, and grow the ML team from there.
How we work
Engineers here work across the stack. Siloed circuit, device, and package engineers do not thrive at Prahelika; you will be expected to understand the neighboring disciplines well enough to argue with them.
AI is a cornerstone of how we engineer. You will build your own workflows with tools such as Claude, ChatGPT, or Codex, compatible with Prahelika's AI ecosystem, with physically grounded checks on what they produce.
No one is asked to cut a corner they would be ashamed to defend. If a deadline can only be met by skipping a reliability test, the deadline moves, and the engineer who refuses to sign off on something unsound is protected, not punished.
Credit and authorship follow contribution, not seniority. If your name belongs on the work, it is there.
Excellence of mind and body is a working condition, not a reward for finishing. Tired decisions end up in hardware; we expect you to sleep, train, and step away, and we do not mistake hours logged for work delivered.
Compensation and benefits
$180K – $250K base salary plus founding-team equity. Final compensation depends on experience and skills.
Full medical, dental, and vision coverage with HSA; 401(k) with 4% company match; daily meals.
How to apply
Apply using the form for this role. If available, please include a brief description of one project that best represents your work on this kind of problem; we read that more closely than a cover letter.
Prahelika is an equal opportunity employer. We hire on demonstrated ability, without regard to any protected characteristic.