Notes from the
edge of the model.
Field notes on what actually breaks in production — agents, retrieval, evaluation, MLOps, and the career decisions nobody writes down. Longer arguments become papers; these are the rest.
353 posts
India has 1 doctor per 1,500 people. AI isn't optional for Indian healthcare — it's existential.
The Indian healthcare AI market: $3B+ by 2027, growing 40%+ annually. Funded by both government initiatives and massive private investment. What's being built: AI diagnostic tools for…
By Pranay Mahendrakar Read postHow companies structure AI teams has changed dramatically.
Old structure (2020-2023): centralized AI team that serves the whole company. Data scientists build models. Engineers deploy them. Long backlogs. Slow delivery. New structure (2026)…
By Pranay Mahendrakar Read postGoogle DeepMind's GNoME discovered 2.2 million new crystal structures using AI.
AI for materials science is creating a job market at the intersection of ML and chemistry/physics that barely existed 3 years ago. Applications: discovering new battery materials for…
By Pranay Mahendrakar Read postIf I were starting an AI company today — with everything I know from building 250+ systems…
Week 1-2: Talk to 50 potential customers in a specific vertical. Not "people who might use AI" — people with a specific expensive problem. Listen more than you pitch. Week 3-4: Build the…
By Pranay Mahendrakar Read postAI testing and evaluation is becoming its own industry.
The problem: every company deploying AI needs to know if it works correctly. But testing AI is fundamentally different from testing software. A model can pass all your tests and still…
By Pranay Mahendrakar Read postClose your eyes and imagine your AI career in 2030. What does it look like?
Here's my vision based on every trend I can see: You probably don't "train models" anymore. Foundation models are commoditized. You assemble, fine-tune, evaluate, and orchestrate them…
By Pranay Mahendrakar Read postLogistics is a $10 trillion global industry.
Route optimization with ML saves delivery companies 15-25% on fuel costs. Demand prediction tells warehouses what to stock before orders arrive. Computer vision automates package sorting…
By Pranay Mahendrakar Read postThe next frontier in conversational AI isn't better language. It's emotional intelligence.
AI systems that detect frustration in a customer's voice and adjust their tone. Chatbots that recognize when someone is confused (not just what they're asking, but HOW they're asking)…
By Pranay Mahendrakar Read postCompliance is a $50B industry that's 90% manual and 100% tedious.
Financial compliance: AI that monitors transactions for regulatory violations in real-time. Healthcare compliance: AI that ensures clinical documentation meets HIPAA standards. Corporate…
By Pranay Mahendrakar Read postAI-generated music is a $3B+ market and growing.
What AI does in music now: generates complete compositions from text descriptions. Creates background music for content at a fraction of traditional costs. Masters and mixes tracks…
By Pranay Mahendrakar Read postThe AI talent war is the most intense hiring competition in tech history.
The numbers: estimated 1M+ unfilled AI positions globally. Top AI engineers receive 5-10 recruiting messages per week. Counter-offers of 30-50% salary increases are common. How to…
By Pranay Mahendrakar Read postI listen to 8-10 hours of AI podcasts weekly.
Consistent themes from leaders at Anthropic, Google, Meta, and top startups: The biggest hiring gap isn't model builders — it's evaluation and deployment engineers. Everyone can train a…
By Pranay Mahendrakar Read post