The Verification Layer.
Taqeem replaces gut-feel hiring with a structured, AI-evaluated engineering verification pipeline. Real code in native Git environments. Anti-cheat telemetry that respects candidate privacy. Transparent scores with no black boxes.
Build, then defend
Build a work sample, then defend it in a grounded interview. We run what the candidate built, then ask them to explain it — because a score on its own cannot tell you who wrote the code.
- 01~30 sec
CV Parsing
Structured extraction of experience, education, and skills from the candidate's uploaded CV. Used for a consistency check against the code evidence — it carries no score of its own.
- 02~2 min
GitHub Analysis
Static analysis of public repositories: commit frequency, code complexity, language diversity, test coverage signals, and architectural patterns. Optional parallel stage — empty GitHub does not block progression, and an unverified account is excluded from the score.
- 03~2 hrs
Build — Coding Task
A proctored, timed coding task in a native Git repository. Candidates use their own tools. Today the submission is scored by static review plus the run log the candidate produces; a hidden black-box test suite is not executed yet, and every report says so on its coding row.
- 04~30 min
Defend — Interview
A timed, text-based interview grounded in the code you submitted — questions on your design decisions, failure modes, a failed test, and one change request. Runs in fullscreen with webcam integrity; three tab or fullscreen exits end the session.
- 05~1 min
Synthesis & Badge
Multi-judge AI weighs the stage results, then level gates decide whether the evidence supports the level the average suggests. The Verified Engineering Badge is issued with the gates that produced it.
Anti-Cheat Telemetry
Integrity signals run silently throughout every session. We use soft signals — not automatic disqualification — and every badge explicitly discloses what was and wasn't detected.
Webcam presence
Continuous webcam frame sampling during coding and interview stages. Frames are stored as evidence; they are not used to change a level or a verdict.
Stage-aware focus policy
Page Visibility API tracks tab switches. Coding logs sustained away-time and treats brief lookups as context. The interview runs in fullscreen and ends after three tab or fullscreen exits.
Keystroke pacing
The editor reports paste events. A stage whose answer arrives as one large paste is scored on what the candidate can defend, not on what was pasted.
AI copilot detection
The judge flags code patterns that look machine-generated. It is an inference from the submitted code, not surveillance of the candidate’s tools — and the report labels it that way. Stated AI use is recorded as a fact with no penalty; only an admission of hiding it touches the integrity gate.
No video is stored. No biometric data is processed. Integrity signals are used only for badge transparency — not automatic disqualification.
AI Evaluation Dimensions
Multiple AI judges independently evaluate each artifact. Scores are aggregated using a weighted synthesis that accounts for signal confidence — so empty GitHub (when skipped) doesn't drag down the final badge.
Code Quality
Naming, structure, test coverage, error handling
Architecture
Modularity, separation of concerns, scalability
Problem Solving
Correctness, edge case handling, debugging approach
Communication
Interview clarity, code review depth, documentation
Seniority
Calibrated against industry benchmarks per track
Sample Badge Breakdown
Overall: 74 / 100 · Senior · GitHub: skipped
For Hiring Teams
The company portal gives you the evidence report: the code they submitted, stage-by-stage score math, the defence transcript, and the integrity facts — including the parts that say “not measured”.
See company plansFilter by verified track
Backend, Frontend, Fullstack, ML/Data. Filter by seniority and score range.
Full evaluation reports
See the candidate's code, AI analysis, interview transcript, and integrity signals.
Direct contact requests
No intermediary. Send a contact request directly — candidates accept or decline.
No agency markup
Flat team fee. Zero placement commission when you hire.
Start today.
Verification is invite-only at launch. Companies pay to contact and hire. No setup fees.