We replace the first-round technical screen with something you can actually check
InterviewLM is a technical assessment platform. Candidates solve repository-shaped problems in an isolated cloud sandbox with an AI assistant they are encouraged to use, and the platform scores how they worked — their problem-solving, their code, and how well they collaborated with the AI — with the session evidence attached to every score.
It is software we build and operate ourselves: a multi-tenant SaaS product that customers sign up for and use directly, currently in production and running ~200 assessments a month.
The first round is the broken part of technical hiring
Every team we talk to has the same four complaints about the screen that sits between an application and an onsite.
First-round screens are slow
A senior engineer spends 45–60 minutes per phone screen, plus scheduling overhead, and most of those hours are spent on candidates who will not advance. Coordinating a single round across timezones routinely adds a week to the funnel.
They are inconsistent
Two interviewers ask different questions, weight answers differently, and write up their notes from memory. The same candidate can pass with one and fail with the other, and nobody can point to the evidence that decided it.
They are resume-driven
When there is no reliable signal from the screen itself, decisions fall back to the resume — company names, school, keyword overlap. That filters out capable engineers with unfamiliar backgrounds and lets polished resumes through unchecked.
And they test the wrong skill
Legacy assessment platforms ban AI and time-box an algorithm puzzle. Every engineer you hire will use AI daily on the actual job, so a test that forbids it measures memorisation, not competence.
Our target audience
We sell to engineering hiring managers and talent acquisition teams at startups and mid-size technology companies — typically companies of 20 to 500 employees that are hiring engineers continuously, have no dedicated assessment team, and cannot afford to spend senior engineering hours on first-round screens.
The engineering hiring manager
Owns the bar and the headcount. They are protecting their team's time, and they need to trust the signal enough to skip a round. They care that scores are defensible and that they can watch the session when they disagree.
The talent acquisition lead
Owns throughput, candidate experience, and time-to-hire. They need rounds that run without scheduling, a funnel that moves at nights and weekends, and a process that survives a fairness review.
We also serve engineering candidates directly: anyone can take a sample assessment and see their own evidence-backed report, and verified candidates can be surfaced to hiring teams through our talent marketplace.
One configured round, run identically for every candidate
Configure the round once
A hiring manager picks the role, the seniority bar, and which rounds to run from 13 assessment types. That configuration becomes the rubric every candidate for that role is scored against — so the round is structured by construction, not by whoever happens to run it.
Candidates work in a real sandbox — with AI
Each candidate gets an isolated container with a real terminal, filesystem, and an AI assistant they are told to use. The task is repository-shaped: debug a failing service, refactor a module, review a pull request. It looks like a Tuesday at work.
We score how they worked, with the evidence attached
Every prompt, edit, command, and test run lands in an append-only event log. Scores are produced per dimension — including how well the candidate collaborated with AI — and each one cites the moments that justify it, so a hiring manager can check any score rather than trust it.
The funnel moves without a calendar
Candidates start the moment they are invited, at any hour, and advance, stall, or route to manual review on rules the team set. Engineers only spend time on candidates who already cleared a structured, evidenced bar.
What we believe
AI is the job, not the cheat
We assume every engineer uses AI daily, and we measure whether they use it well.
Assessments should feel like work
Real repositories and real terminals, not artificial puzzles built to be graded automatically.
Evidence over vibes
Every score cites the session moments behind it, and says so when the evidence is thin.
Decisions, not dashboards
The output is a recommendation a hiring manager can act on or argue with — not a report that collects dust.
Who builds this
InterviewLM is a solo-founded company. One engineer builds and operates the platform end to end — product, backend, agents, and infrastructure.
Ragavendra Selvam S
Founder & CEO
Bengaluru, India · 9+ years in software engineering
Ragavendra spent nine years building and running production backend systems before starting InterviewLM — at Amazon, FourKites, Acceldata, Pillow, and most recently Groupon, where he reached SDE 4 and owned the platform team's proxy, orchestrator, and user services.
He built InterviewLM after years on both sides of the technical screen: interviewing candidates against algorithm puzzles that predicted almost nothing about how they would perform, while doing a day job where the difference between a good engineer and a great one increasingly came down to how well they worked with AI tooling. The platform is the assessment he wanted as a hiring manager — structured, evidenced, and honest about what it can and cannot measure.
Experience
Owned the proxy, orchestrator, and user services for the platform team. Built a fraud-prevention system protecting 800M+ users from password-based attacks, and cut p99 latency 25% while taking error rates from 3% to under 1%.
Built the entire backend for a consumer crypto product from zero — payments, balances, interest accrual, and multi-chain withdrawals — and led the team delivering it.
Wrote a lightweight Go agent collecting thousands of host metrics per second for real-time data observability.
Built a route-planning recommendation engine on Neo4j and a network-optimisation platform for supply-chain analytics.
Migrated a monolithic payments system to microservices, designing the backfill and reconciliation strategy behind the cutover.
Company details
InterviewLM is the product of Corrirrus Innovations OPC Pvt Ltd, a technology company registered in India. We build and operate one proprietary software platform and sell access to it as a subscription and usage-based product. We are not an agency, a consultancy, or a staffing firm — we do not sell services, resell others' software, or place candidates for a fee.