Real, anonymous reports from people who interviewed for Software Engineer at Speechify, newest first and distilled into what to expect across the loop.
My process started with a single, long technical assessment that was basically all about debugging with an LLM. After an initial step, I was dropped into a 90-minute proctored-style codebase exercise where I could use AI tools while working in an unfamiliar repository. The goal wasn’t just to make the tests pass—it was to audit and improve things like security, performance bottlenecks, and observability.
After a recruiter call that covered my background and walked through what the technical rounds would look like, I ended up doing a pretty full set of engineering screens over the next stretch of time. The first technical hurdle felt very structured: a timed LLM refactoring test built around fixing problems in a full-stack app / codebase. It was centered on using the AI-assisted workflow but also on doing the “real engineer” parts—writing or keeping notes, and dealing with the repo as it existed instead of rewriting everything.
Candidates often start with an automated email directing them to a timed technical assessment, which can feel impersonal and jarring without prior human interaction. Many reported feeling unprepared due to the lack of context about the role before diving into assessments.
The technical assessments are typically proctored, requiring candidates to use their camera and screen sharing, which some found invasive and stressful. The assessments often focus on debugging and parsing tasks within a strict time limit, leading to a high-pressure environment.
Candidates experience live coding sessions that often involve fixing issues in a codebase or implementing features under observation, which can limit interaction and make it challenging to demonstrate problem-solving skills. The expectation to perform under tight constraints is a common theme.
The interview process includes system design discussions that are generally open-ended and can focus on real-time AI applications, allowing candidates to showcase their design thinking. These rounds are often seen as less structured than coding assessments.
A recurring concern is the absence of feedback after interviews, leaving candidates uncertain about their performance and areas for improvement. Many expressed frustration over not receiving closure or constructive insights post-assessment.
Candidates noted that the overall experience varied significantly, with some describing it as organized and coherent, while others felt it was chaotic and poorly communicated. The final stages often included conversations with leadership, which helped clarify fit and motivations.