Most hardware and embedded projects that go badly go badly because of who got hired, not because the engineering problem was unusually hard. The technical work is rarely the surprise. The vendor is.
This covers how to actually find a good firm and what to ask before you sign anything. For contract terms, red flags during execution and what to do when a project starts going sideways, see Red Flags When Hiring Engineering Firms. This post is about picking well in the first place.
Where to actually find firms
Referral sources aren’t equal.
Best: engineering managers at similar companies who’ve actually worked with the firm. Component vendor FAEs know too. They see dozens of designs a month and know who’s competent. So do certification lab engineers. They see quality across hundreds of submissions, and they know exactly which firms show up clean and which ones repeat the same rookie mistakes.
Worth trying: industry-specific associations (AAMI for medical, SAE for automotive) tend to beat generic directories like the IEEE consultant list. GitHub can show you actual firmware quality if a firm has public work.
Skip: freelance marketplaces, generic business directories, cold outreach. Firms that get real referral business don’t usually need to advertise on Upwork.
When someone gives you a name, push past “they were great to work with.” Ask what would’ve made the engagement better. Ask whether they’d hand the firm something harder next time. That second question is the real signal. It tells you whether the referrer trusts the firm’s ceiling, not just their floor.
Aim for 5-8 candidates before you start screening: a couple from referrals, a couple from vendor/lab recommendations, one or two you found yourself.
Screen by email before you schedule a single call
A surprising number of firms can’t clear this bar. Ask directly:
“Show me three products you’ve designed that are currently shipping.” Not prototypes. Production, with part numbers if they’ll share them. A firm that can only point to demos, or whose examples are all several years old, is telling you something. New firms should be able to point to what the founders shipped at previous jobs instead.
“What’s your typical timeline for something like ours?” You’re not looking for a number, you’re looking for whether they can break the work into phases with realistic buffer. A timeline meaningfully shorter than everyone else’s is a red flag, not a selling point.
“Who would actually work on this?” If the answer is a project manager and a vague “our experienced team,” that’s the answer. You want names, backgrounds and direct access to the people doing the work, not a layer between you and the engineer.
“What do we own at the end?” Complete files, source code and documentation in standard formats, no ongoing licensing, no vendor lock-in. Anything more complicated than that is worth reading twice.
Once you’re past screening: ask real technical questions
This is where salespeople and engineers stop sounding the same.
Ask them to walk through their design process for a specific piece of your project. A competent engineer gets concrete fast: manufacturing implications, regulatory considerations, real trade-offs. A generic answer at this stage is usually a generic answer for the whole engagement.
Ask what they think the hardest part of your project will be. Firms that immediately spot real risk have actually thought about it. “Should be pretty straightforward” from someone who hasn’t built your product before is not reassuring, it’s a tell.
Ask how they test and validate designs, and listen for whether they mention EMC pre-compliance, environmental testing or structured verification, versus “we’ll test it once it’s built.” And ask what would make them turn a project down or hand it to someone else. Firms that have never said no to anything usually haven’t been doing this long enough to need to.
The honest version of firm size
Established firms bring process and depth, but plenty of them are coasting on a reputation built years ago while junior staff does the actual work now. Newer firms bring direct access to whoever founded the place and real motivation, but some of them are just replicating whatever their last employer did, badly, at a lower price.
Neither age nor size is the signal. Current competence is. Verify it directly instead of trusting the firm’s own pitch about how long they’ve been around.
If you’re still not sure
A small paid trial engagement (an architecture review, a proof-of-concept for the riskiest subsystem, a few days of real work) tells you more than another round of calls. Pay market rate for it (never “audition for free”), keep IP terms standard, and treat it as a real decision point, not a formality.
If nothing you’ve seen gives you confidence, the right move is to keep looking, not to pick the least-bad option under deadline pressure. A bad hire here costs a lot more than the extra week of searching.
Want to see how we handle scoping and contracting? Start with our services page.