Vivek Ravisankar

Vivek Ravisankar is HackerRank’s co-founder and CEO. His AI-relevant work includes public analysis of coding models and explanations of how HackerRank assesses developers who build with AI. In his June 2026 account of Orchestrate, he emphasized a combination of working software, observable implementation, deliberate AI collaboration, and the builder’s ability to explain the system.[2][4]

In his June 8, 2026 article, Ravisankar explained Orchestrate, HackerRank’s 24-hour AI-agent hackathon. Participants built a terminal-based support-triage agent for tickets spanning HackerRank, Anthropic, and Visa, using 774 Markdown documents as its only knowledge base. The 29-ticket bank included prompt injection and jailbreaking attempts. Agents had to classify requests, assess urgency, and justify replying or escalating; choosing either action for every ticket was invalid. Participants chose their own retrieval architecture, models, frameworks, prompts, and programming language.[4]

The team evaluated four artifacts: source code, ticket outputs, AI chat transcripts, and a 30-minute voice interview with an AI judge. These tested different aspects of the work: what was implemented, whether it worked, how the participant directed AI during development, and whether they understood the resulting system. The code rubric rejected unsupported descriptions: “We reward only what is observable in the source code. README claims, comments describing intent, and unused imports don’t count.”[4]

Ravisankar’s account described a consequential adjustment for the first edition: chat transcripts received 10% of the final score, following upload errors and signs of manipulation. Code, outputs, and interviews each received 30%. The interview probed architecture, failure modes, testing, production viability, and which AI suggestions the participant had accepted or challenged.[4]

HackerRank reported that, among 1,349 participants who submitted all four required artifacts, no pair of evaluation signals had a Spearman rank correlation above 0.45. Ravisankar interpreted this as evidence that producing code, obtaining correct outputs, and defending design choices captured related but distinct skills. He also cautioned against reading Claude Code’s prevalence among highly ranked submissions as proof of tool superiority: his interpretation emphasized participants’ planning, testing, iteration, and direction of the AI tool.[4]

Before Orchestrate, Ravisankar framed AI’s role through changes in how developers express and organize their work. In April 2023, he argued that AI extended two trends: programming’s convergence with natural language and movement toward higher levels of abstraction. He expected AI assistants to take over time-consuming tasks and make development and personalized learning more accessible. His prediction that AI would change development without replacing developers came with an explicit qualification: “Just remember that probably does not equal certainly.”[6]

In April 2025, his analysis mapped model capabilities across planning, design, implementation, testing, deployment, and maintenance, drawing on SWE-Bench-verified, SWE-Lancer, LiveCodeBench, and HackerRank-ASTRA. He emphasized reliability alongside raw performance, describing ASTRA’s assessment of consistency across 32 runs. His conclusion at that time stressed limitations in complex multi-file projects, deep contextual understanding, and architectural decisions, and favored models as collaborative tools rather than autonomous controllers of development.[5]

Ravisankar’s earlier career supplies context for his work on developer assessment. In a Mixergy interview published November 19, 2014, he recalled meeting co-founder Hari Karunanidhi while studying computer science. The two taught Linux classes together, which he said paid more than half of their college fees. Ravisankar later left Amazon, while Hari—listed by HackerRank under his full name, Harishankaran Karunanidhi—left IBM to start their business.[7][2]

Their initial InterviewStreet service charged students for mock interviews with employees of prospective employers. Ravisankar recalled difficult payment collection, thin margins, scheduling conflicts, and weak incentives for interviewers. After roughly two years, the founders shifted toward helping companies screen programmers through coding challenges with automated evaluation. He named BankBazaar as the first paying customer. The early product still required customers to call or email either founder for candidate scores, which the founders assembled and sent manually.[7]

Ravisankar also described a personal hiring mistake: he had overvalued an unnamed candidate’s previous employers and overridden doubts about how that person would perform with fewer resources. He said the hire failed for the reasons he had initially questioned. His stated mission in the same interview was hiring where “the only thing that matters is your skill”—rather than school, previous employer, or country.[7]