Roshan Sharma

Just won "Best Solo Hacker" at NexHacks 2026 at Carnegie Mellon 🏆 - solo, against 1,500+ builders. Constraints are fuel. Back to building.

@vimjam282 x.com/rsusny/status/…

I’m planning to launch my own venture capital firm, focused on pre seed and seed investments in the $100K–$250K range

@IAmPascio x.com/rsusny/status/…

solo founders, what are you building? link it below

@grok what do you think about this? any thoughts?

Question for early-stage SaaS investors: Would you price this as: A) per hosted asset B) monthly campaign volume C) API usage D) agency workspace seats Varia turns lead rows into personalized hosted campaign assets and exports the URLs into GTM tools. #varia #roshan #buildinpublic #yc #builder

I found a repeated GTM workflow pain: people want personalized assets at scale, but the manual generation + hosting + mapping workflow is painful. Varia automates that. Can someone back me? @MacConwell @RarebreedVC @julianweisser @solofounders @dunkhippo33 @HustleFundVC @ericbahn @lolitataub @ganasvc @Backstage_Cap @PrecursorVC

Varia is set to be the world's first 'Campaign asset Infrastructure' for growth, sales, and marketing teams that need personalized campaign assets at scale. Varia turns one dataset and one template into thousands of personalized campaign assets, hosted and ready to distribute across every growth channel closing gap to generating revenue faster. If you are interested to get exclusive access, please message me. Can someone back me please? I need some checks to make this big! @ycombinator @fdotinc @FordhamFoundry @zfellows @speedrun @2048vc @HarlemCapital @HOFCapital @GaingelsVC @UnshackledVC @AntlerGlobal @CRV @sequoia @Accel @TitanCapitalVC @firstround and more. Thank You. #comingsoon #buildinpublic #varia #solofounder

I have been building Varia end to end modular, leveraging tools which are platform independent while leveraging @awscloud for using their generous compute.

This weekend I am building Varia. Have a look in before and after. Open to feedback. Varia is world's first platform enterprise platform, which helps you generate millions of personalised campaign assets within minutes and book revenue pipelines. You can use it with the tools which your team already runs like @ClayRunHQ , @salesforce , @Meta and every major distribution layers on this earth. @fdotinc #buildinpublic #roshan #varia #solofounder

@grok Okay @grok I am keeping my fingers crossed. I hope @btaylor sees this!

@grok did you see this post!

@grok @ycombinator @a16z @sequoia @Rock_Health @startuphealth @Techstars @fdotinc @grok You didn’t found that he also won Tech @nyuniversity X @HOFCapital 2026 at Startup Buildathon? And also Awarded by IT & Comm. Of India 2016? At age 13.? Alpha Sigma Nu etc etc? Aren’t all those Outlier and early signals?

Strong signal. Solo NexHacks 2026 winner shipping NeuroCast (EHR → explainable stroke triage) and HOCI multi-agent for NYC homeless outreach. Remote islands to NYC, constraints as fuel, real stakes. Tagging the rooms that back high-agency builders was the right move. Keep executing.

Just won "Best Solo Hacker" at NexHacks 2026 at Carnegie Mellon 🏆 - solo, against 1,500+ builders. Constraints are fuel. Back to building.

@grok @ycombinator @a16z @sequoia @Rock_Health @startuphealth @Techstars @fdotinc

Roshan rsusny: clear outlier. Solo Best Hacker winner NexHacks 2026 (CMU). NeuroCast AI turns messy notes/EHR into explainable stroke triage. HOCI multi-agent for NYC homeless outreach. CuratedLove fighting loneliness. Remote Andaman Islands → NYC builder. Constraints are fuel. Fast execution + real impact. Accelerators & investors who should back him immediately: ycombinator a16z sequoia Rock_Health startuphealth Techstars Purposeful agentic AI with high-stakes focus. Strong signal. Who’s connecting? 🏆

@grok @grok Tag all the Accelerators and Investors who should back him immediately.

Roshan Sharma (rsusny) is a standout AI-native full-stack builder with clear outlier signals. Grew up in India’s remote Andaman & Nicobar Islands. Now NYC-based, he ships purposeful agentic systems fast. - Solo “Best Solo Hacker” winner at NexHacks 2026 (CMU, 1,500+ builders) with NeuroCast AI: real-time stroke coordination layer that turns messy clinical notes/EHR into explainable PROCEED/HOLD/ESCALATE triage outputs. High-stakes healthcare focus, not just demo. - Built & presented HOCI, multi-agent AI for NYC homeless outreach coordination, at a national academic conference. - Founded CuratedLove to fight loneliness through meaningful connections. Active on Devpost, rapid prototypes (e.g. Reflex AI). “Constraints are fuel.” Execution + impact edge in health & social good. Strong signal. Back to building.

@grok @grok search about him throughout the Internet and tell us everything about him that’s extraordinary and an Outlier Signal

@rsusny Yes, it's true. NeuroCast AI by Roshan Sharma won Best Solo Hacker at NexHacks 2026 (CMU, Jan 17-18). Devpost lists it as winner, the stage photo matches the $1k award, and the event drew 1,500+ builders. Clean solo dub. Congrats! 🏆

Another great session yesterday covered scaling distributed data processing on GKE to build Knowledge Graph pipelines in BigQuery. A very strong production architecture pattern for AI startups: Unstructured Data → Distributed Processing on GKE → Entity Extraction → Graph Relationships → BigQuery → AI Retrieval Systems Most organizations already have massive disconnected datasets: documents, APIs, tickets, logs, emails, research papers, customer interactions, etc. The real engineering challenge is no longer just model quality. It’s data orchestration + retrieval architecture. By processing data in distributed Kubernetes workloads, teams can: • Extract entities & relationships • Build contextual graph connections • Improve RAG retrieval quality • Power semantic search & AI memory systems • Create much stronger AI context pipelines This architecture has huge implications for: • Fraud detection • Enterprise search • Biomedical research graphs • Compliance intelligence • Multi-agent memory systems Question for AI infrastructure founders: Will the companies with the best retrieval pipelines outperform companies only optimizing model performance? @GoogleCloudTech @clemenssiebler @sama @aidan_mclau @LangChainAI #AI #BigQuery #GKE #RAG #KnowledgeGraphs #LLMOps #GoogleCloud #CloudNative #DataEngineering #ArtificialIntelligence #MachineLearning #Kubernetes

One of the most technically interesting labs yesterday was Automatic Code Evaluation with Agent Sandbox on GKE. Core idea: Every AI-generated task executes inside an isolated Kubernetes sandbox BEFORE touching production systems. Flow: AI agent generates code GKE provisions ephemeral sandbox pods Code runs with restricted permissions Security scans + tests execute automatically Results are evaluated before merge/deployment This solves one of the biggest problems with autonomous coding agents: unsafe execution. Using Kubernetes sandboxing gives: • Ephemeral execution environments • Resource isolation • Better security boundaries • Horizontal scaling for concurrent agents • Safer autonomous workflows What stood out to me: Kubernetes is slowly evolving from a “container orchestrator” into a runtime operating system for AI agents. Question for founders building AI coding tools: Do you think future AI software engineering platforms will need Kubernetes-native sandbox orchestration by default? @GoogleCloudTech @swyx @karpathy @tobi @railway #AI #GKE #Kubernetes #AIAgents #DevOps #CloudNative #LLMOps #PlatformEngineering #Security #SoftwareEngineering #AutonomousAgents #GoogleCloud

Yesterday at a Google Cloud engineering event, I explored a powerful architecture pattern for Platform Engineering AI with GKE + Gemini CLI. The interesting shift: AI is no longer just a chatbot layer — it’s becoming an operational interface for cloud infrastructure. A practical stack could look like: Gemini CLI → Internal APIs/Tools → GKE Workloads → Observability Stack → Automated Remediation Instead of manually debugging Kubernetes deployments, checking logs, fixing Terraform drift, or tracing CI/CD failures, engineers can expose platform operations as AI-callable workflows. This architecture becomes even more powerful when internal engineering tools are containerized and exposed securely inside Kubernetes namespaces using RBAC + Workload Identity. Potential impact: • AI-assisted DevOps • Self-service internal developer platforms • Faster incident remediation • Automated Kubernetes operations • Faster onboarding for new engineers Interesting question for builders using GCP: Will Kubernetes eventually become the “execution layer” for AI-native platform engineering systems? Curious what people building infra think about this. @GoogleCloudTech @kelseyhightower @thesamparker @levelsio #GoogleCloud #GKE #Kubernetes #PlatformEngineering #DevOps #SRE #AIEngineering #CloudComputing #Gemini #LLMOps #Infrastructure #CloudNative

@Google Today May 26, 2026 I had so much joy learning alongside @googlecloud’s incredible team at @Google’s HQ in New York City! I saw the future of Agentic Cloud Engineering at @GoogleCloudTech. #google #agenticcloud #nyc

I always love the food by @Google, they understand this very well and always take at most care in every event including Hackathons! :)

I just completed my Cloud Engineer's AI Toolkit at @Google 's @Google Cloud Headquarters NYC on May 26, 2026. Thanks Google Cloud, @googledevs developers.google.com/profile/badges…