Talent, Tech & Construction | Stream 99
This stream transcript covers two main conversations: Vivek from HackerRank discussing how technical hiring is evolving in the AI era, focusing on AI fluency over code correctness, and Kanna from MeltPlan explaining how AI is being applied to fix the deeply fragmented and inefficient construction planning industry.
Summary
The first segment features Vivek from HackerRank discussing the fundamental transformation in technical hiring driven by AI. He argues that traditional coding assessments focused on converting English specs into code are obsolete since AI can do that better. HackerRank is shifting to evaluating 'AI fluency' — how effectively candidates use AI agents to accomplish real tasks — through a 'Plan, Build, Review' framework conducted in agent developer environments rather than IDEs. Their AI interviewer, Chakra, assesses candidates on thought process, trade-off judgment, and code reasoning rather than mere correctness.
Vivek introduces the concept that every IC (individual contributor) must now think of themselves as an 'engineering manager of agents,' applying to AI agents the same task delegation, performance feedback, and resource allocation skills traditionally applied to human teams. He distinguishes between AI skills (prompt engineering, RAG, fine-tuning) and AI fluency (using agents effectively to solve customer problems), warning that companies often conflate the two when hiring. He also observes that junior developers are naturally AI-native but lack judgment, while senior developers have strong judgment but must overcome muscle memory — suggesting the two groups have much to learn from each other.
Vivek also notes that despite social media narratives claiming software engineering is dying, HackerRank's data shows new grad hiring assessments are up 25-30% year over year. He observes that engineering skills are spreading across all business functions, with customer support, demand gen, and account management roles now requiring technical competency. HackerRank holds the distinction of being the first India-based company accepted into Y Combinator (2011).
The second segment features Kanna from MeltPlan, who is building an AI-powered pre-construction planning platform. He explains that construction is the world's second-largest industry but has seen declining productivity for four decades, attributing this to extreme fragmentation — where tasks once managed by a single master builder are now spread across 35-40 specialized firms. He uses the Empire State Building (built in 19 months in 1931) versus Salesforce Tower (7 years) as a striking example of how fragmentation has degraded efficiency. MeltPlan aims to serve architects, engineers, and contractors on a shared platform, giving each party visibility into the others' plans. Kanna argues that traditional SaaS failed construction because it required users to manually input data from documents and images, whereas AI can now extract intelligence directly from those artifacts. He believes better planning will reduce waste, make housing more affordable, enable more construction projects, and unlock the potential of prefabrication.
About this episode
<p>Two founders joining us today:</p><p>Vivek Ravisankar (HackerRank) on whether tech hiring is actually fixed, what AI does to the job of being an engineer, and where the human sits when machines write most of the code.</p><p>Kanav Hasija (MeltPlan) on why construction productivity has been falling for 50 years, what pre-construction planning has to do with everything that goes wrong on a build, and why Bessemer backed him at less than a year old.</p><p><br /></p>
Key Insights
- Vivek argues that evaluating developers on converting English specs into code is now obsolete, since AI can perform that task better than humans, and the new evaluation focus should be on judgment, trade-offs, and AI fluency.
- Vivek claims that every IC must now think of themselves as an engineering manager of agents — performing the same task delegation, context-setting, and output review functions for AI that managers previously did for human teams.
- Vivek observes a paradox where engineering managers are now being evaluated on whether they can be hands-on ICs, while ICs are being evaluated on whether they can manage agents like engineering managers.
- Vivek states that HackerRank's data contradicts social media narratives: new grad hiring assessments are up 25-30% year over year, though the skills being sought have shifted significantly toward AI fluency.
- Vivek argues that large-scale IT services firms like TCS and Accenture are actually more progressive on AI-native hiring than many tech-forward organizations, likely because they must signal AI capability to enterprise clients when staffing teams.
- Kanna argues that the Empire State Building was completed in 19 months in 1931 — including planning — because a single master builder controlled the entire process, and that modern fragmentation across 35+ specialized firms is the primary reason productivity in construction has declined for four decades.
- Kanna claims that traditional SaaS fundamentally failed construction because it required workers to manually enter data from documents and images into forms, creating more work than value, whereas AI can now extract intelligence directly from those source materials.
- Kanna argues that improved pre-construction planning will unlock prefabrication at scale, because prefab's main barrier is not cost but the requirement to commit to fully complete plans early — a requirement that better AI planning tools can make feasible.
Topics
Transcript
The second is what are you evaluating? It's no longer about code correctness, it's about AI fluency and your software engineering fundamentals matter much more. Your critical thinking, judgment, when do you do the right trade-offs matter much more. The third one is a candidate expedience. How do you go from an IDE to an agent developer environment? Historically, pretty much most companies ask the code-style question, evaluate correctness, and put them on an IDE. We're changing that to real tasks, evaluating on their AI fluency on an agent-developer environment. So that's what we're doing on the first part. I can talk endlessly on the second part. I can also talk endlessly on the first part as well on the…
Full transcript available for MurmurCast members
Sign Up to AccessMore from The Offline Network
India’s Startup Titans on Building, Surviving & Winning | Stream 100
This is the 100th live stream episode of T1, featuring founders and executives from five Indian startups: Servify (device lifecycle management), Square Yards (proptech), UpGrad (edtech), Shadowfax (logistics), and InMobi (adtech/consumer internet). Each guest discusses their company's business model, growth journey, competitive strategy, and views on AI and the future of their respective industries.
Community, E-commerce & Starting Over | Episode 98
Stream 98 features Ishan Rakshath discussing his e-commerce infrastructure companies Shopflow and Sortment, followed by a four-way conversation with Suhas from ProductFolks about AI's impact on product roles, community building, and the evolving startup ecosystem between India and the US.
Community, E-commerce & Starting Over | Episode 98
Episode 98 (mislabeled as 97 in the intro) features three guest segments: Sahil Shah of TM Systems discussing OT cybersecurity and AI threats, Dhaval Jain of Mave presenting a consumer brain stimulation headset, and Afshan and Vinay of Legend of Toys explaining their full-stack RC car brand with IP-building ambitions.