Launched this week
Lytmus
Interview tracking + AI coach for software engineers
9 followers
Interview tracking + AI coach for software engineers
9 followers
Most job trackers stop at applied → interviewing → offer. Lytmus goes deeper: log every round, every question asked, your approach, and a confidence score (1–5). Then it surfaces patterns — which topics tank your loops, where confidence and outcomes diverge, which companies grind you down. An AI coach reads your real interview history (not generic advice) and drills the exact gaps. Built by an engineer mid-search. Free up to 10 applications.
Hey Product Hunt — Anuj here, maker of Lytmus.
I built this during my own job search, out of frustration with every other tracker I tried.
Here's what kept happening: I'd finish a tough onsite, the questions still fresh — and then dump everything into a Notion doc I'd never reopen. My Huntr board told me what stage I was at, but nothing about what was actually going wrong inside the interviews. Three searches in, I started noticing patterns hiding in my own history that no tool could surface: I lose system design more often when the loop has 4+ rounds. My "confidence 2" answers correlate with "no offer" way more than I'd like to admit. There were lessons buried in my own data, and no tool to extract them.
So I built Lytmus.
What it does
- Kanban board for applications (the table-stakes part)
- Per-round capture: kind (DSA / system design / behavioral / take-home / …),
interviewers, outcome, schedule
- Per-question capture: prompt, approach, solution (CodeMirror for code, Tiptap for
prose), confidence 1–5, difficulty, tags, links
- Notes & attachments scoped to application / round / question, typed by source
(recruiter info, prep, self-reflection, peer tip, interviewer feedback)
- Metrics that mean something: application funnel, tag performance,
confidence-vs-outcome, activity heatmap
- AI coach (Claude-powered) that reads your actual data — your weak tags, low-confidence
questions, past reflections — and drills the real gaps. Not generic LeetCode prompts.
- MCP support if you'd rather log interview rounds, questions using your favorite LLM or pull Lytmus into your own AI workflow
The bet: capture enough structured signal and patterns emerge — and that same data makes an AI coach that actually knows you, not generic LeetCode prompts.