I'm Yash.

Sell the work, not the tool. Evaluate the agent, not the feature.

I ship AI products where success is measured in agent reliability, not feature count. Founding PM at Voltade, where I drive 0-to-1 agent platforms (Studio, Envoy) and the evaluation frameworks behind them. Five years of engineering before that.

100M+ events/day at Ninja Van
230K AI interactions/day across Voltade agents
50K+ users from zero at AfterClass
7.74B tokens through Claude Code →

Track Record

  • VoltadeDefined strategy for agent platform and conversational CRM. Established evaluation frameworks for production agent reliability across 230K+ daily AI interactions.
  • Ninja VanOwned platform products in the CTO’s office. Notification infrastructure, parcel booking, voice calling across 6 markets.
  • AfterClassCo-founded a student platform that’s still active today with 50K+ users.

View Projects About Me Email Me

Press ⌘K to explore.

Featured Projects

Products I’ve built and shipped.
Vobase

Open-source app framework for AI coding agents. Built a Bun + TypeScript stack with sensible defaults, auth, Postgres, jobs, agent runtime, so coding agents like Claude Code generate working code on the first try.

Focus
open source, agent infrastructure, developer experience
View on GitHub
Voltade Studio preview Voltade Studio

No-code AI agent builder. Defined the product vision and shipped a platform where non-technical teams describe what they need in plain English and get working agents deployed across WhatsApp, Telegram, and Web.

Focus
product strategy, AI UX, shipping 0-to-1
Visit
Envoy CRM

Conversational CRM for SMEs. WhatsApp-first inbox where a per-tenant agent triages, drafts, and replies; humans approve. 100+ active SME deployments, 230K+ AI interactions/day, 99.65% success rate.

Focus
AI-native CRM, conversational UX, agent reliability
Read more
Discover - Synthesis Partners

Enterprise AI for strategic intelligence. Drove product direction and engineering for an NLP/RAG platform serving Fortune 500 clients, where analysts explore millions of documents through topic clustering, network analysis, and theme visualisation.

Focus
enterprise AI, RAG, F500 stakeholder management
Visit
View all projects →

Start here

Three posts that capture how I think about AI products: a 0-to-1 case study, an evaluation framework, and the economics of model selection.
· 12 min read
How I Evaluate AI Agents (and Why Most Teams Get It Wrong)

Most agent evals measure the wrong things. After running two agents in production for six months, here's the framework I actually use, with real metrics, LLM-as-judge calibration data, and the $300 lesson that started it all.

 · 6 min read
Cost is a product feature

Notes on when to use Haiku, when Opus is worth it, and why model choice is a product decision wearing engineering clothes.

View all posts →
Shipping pace 109 active days in the last 20 weeks
2025-12-27: level 12025-12-28: level 12025-12-29: level 12025-12-30: level 12025-12-31: level 12026-01-01: level 12026-01-02: level 12026-01-03: level 02026-01-04: level 12026-01-05: level 12026-01-06: level 02026-01-07: level 02026-01-08: level 12026-01-09: level 12026-01-10: level 12026-01-11: level 02026-01-12: level 12026-01-13: level 12026-01-14: level 12026-01-15: level 12026-01-16: level 12026-01-17: level 12026-01-18: level 12026-01-19: level 32026-01-20: level 12026-01-21: level 12026-01-22: level 02026-01-23: level 12026-01-24: level 02026-01-25: level 12026-01-26: level 12026-01-27: level 22026-01-28: level 12026-01-29: level 12026-01-30: level 12026-01-31: level 02026-02-01: level 12026-02-02: level 02026-02-03: level 12026-02-04: level 12026-02-05: level 12026-02-06: level 12026-02-07: level 12026-02-08: level 12026-02-09: level 12026-02-10: level 12026-02-11: level 12026-02-12: level 12026-02-13: level 12026-02-14: level 12026-02-15: level 12026-02-16: level 02026-02-17: level 12026-02-18: level 12026-02-19: level 02026-02-20: level 12026-02-21: level 12026-02-22: level 12026-02-23: level 12026-02-24: level 12026-02-25: level 12026-02-26: level 12026-02-27: level 12026-02-28: level 42026-03-01: level 12026-03-02: level 12026-03-03: level 12026-03-04: level 42026-03-05: level 12026-03-06: level 22026-03-07: level 02026-03-08: level 12026-03-09: level 12026-03-10: level 22026-03-11: level 42026-03-12: level 42026-03-13: level 02026-03-14: level 12026-03-15: level 12026-03-16: level 12026-03-17: level 12026-03-18: level 32026-03-19: level 12026-03-20: level 42026-03-21: level 02026-03-22: level 12026-03-23: level 12026-03-24: level 12026-03-25: level 02026-03-26: level 12026-03-27: level 12026-03-28: level 02026-03-29: level 12026-03-30: level 02026-03-31: level 12026-04-01: level 12026-04-02: level 12026-04-03: level 12026-04-04: level 12026-04-05: level 12026-04-06: level 12026-04-07: level 12026-04-08: level 02026-04-09: level 02026-04-10: level 42026-04-11: level 02026-04-12: level 02026-04-13: level 02026-04-14: level 12026-04-15: level 32026-04-16: level 12026-04-17: level 12026-04-18: level 02026-04-19: level 02026-04-20: level 22026-04-21: level 12026-04-22: level 12026-04-23: level 12026-04-24: level 02026-04-25: level 42026-04-26: level 02026-04-27: level 02026-04-28: level 02026-04-29: level 32026-04-30: level 12026-05-01: level 32026-05-02: level 42026-05-03: level 22026-05-04: level 12026-05-05: level 12026-05-06: level 02026-05-07: level 02026-05-08: level 12026-05-09: level 12026-05-10: level 02026-05-11: level 12026-05-12: level 12026-05-13: level 22026-05-14: level 02026-05-15: level 1 View on GitHub