What We're Building

Candid is building the AI layer for EPCs and owner operators.

We use AI to reduce the manual effort behind preconstruction engineering by automating document understanding, consistency checks, and repetitive workflow steps across major disciplines. The goal is to make preconstruction 10x faster without sacrificing engineering quality.

We recently raised $6.5M from top AI and industrial investors (including Schneider Electric and Meta’s Chief AI Scientist). Our team includes engineers from MIT, Carnegie Mellon, major LNG and power projects, and leading EPCs.

Why This Role Exists

We need someone who has lived inside Pre-FEED and FEED. Someone who knows what a design basis package actually looks like at 3 AM before a gate review, who has directed engineers through datasheet cycles, and who has felt the pain of building the same thing from scratch on every project.

You are not here to manage a backlog. You are here to direct real project delivery for Candid's clients and run the engineering team doing it, while shaping the AI tools that redefine how this work gets done. You are the bridge between deep domain knowledge and a product team building at startup speed.

This Role

This is a player-coach role at a startup, not a director seat at a 5,000-person EPC. You'll have a small, sharp discipline team reporting to you, not three layers and an army. You'll direct delivery and manage the team, and you'll also personally own quality and roll up your sleeves on the work when it counts, especially early. If the last time you touched a deliverable was a decade ago and you only direct through others, this isn't the right fit.

Day to day you'll:

Direct Pre-FEED and FEED engagements for Candid's clients across power generation, gas-to-power, LNG, midstream, and downstream projects

Lead and manage the discipline engineering team (process, electrical, mechanical, piping), set the bar, and own delivery quality end to end

Own the assembly and quality of design basis packages, heat and mass balances, single-lines, equipment lists, P&IDs and PFDs, datasheets, utility summaries, and gate-review packages

Be the authoritative engineering voice in client conversations with project development managers, VPs of Projects, and heads of engineering

Work directly with our product and AI team to turn real Pre-FEED and FEED workflows into automation targets. What takes 600 hours today should take 60 tomorrow

Be the quality gate on Candid's automated outputs: catch what's wrong before it reaches a client, and feed it back so the product gets sharper

Advise on execution strategy, PDRI scoring, and gate-review readiness for client engagements

Shape Candid's delivery methodology and standards as we scale from early clients to a repeatable practice

Who You Are

15+ years in capital project execution, with significant time in power generation, gas-to-power, LNG, midstream, or petrochemical facilities

You have directed or led Pre-FEED, FEED, or Concept Select phases end to end. Directed, not just participated

Direct experience leading power or gas-to-power projects (combined cycle, simple cycle, cogeneration, or gas-to-power for data center and BTM campuses). Multi-sector experience across LNG, midstream, and downstream on top of that is a strong plus, not a requirement to explain away

You've managed multi-discipline engineering teams across process, mechanical, piping, electrical, and instrumentation. You know how to set a quality bar and hold it

Deep fluency in FEL stage-gate processes, PDRI, AACE cost estimate classes, design basis development, and the full document ecosystem that feeds a gate review

Fluent across the codes and frameworks that govern these projects (ASME, API, IEEE, NFPA, and the relevant power and interconnection standards), at the level of someone who directs the work, not necessarily runs every calc

Strong opinions on where engineering hours get wasted, where errors propagate, and what should be automated

Comfortable operating lean, with ambiguity, and without bureaucracy. You'd rather ship and iterate than wait for the perfect process