Building autonomous AI workflows & 0-1 SaaS products.
Bridging the gap between technical architecture and commercial outcomes through API orchestration, LLM reasoning constraints, and outcome-led growth systems.
Tech Stack & Skills
Orchestration & Backend
AI Models & Synthesis
Data & APIs
Product Ops & Execution
How I Build
Outcome over Specs
I don't just write strategy decks; I build the API logic and ship the MVP.
Serverless Orchestration
Using n8n, Supabase, and custom Python to bypass traditional engineering bottlenecks.
Data-Driven GTM
Building autonomous distribution engines that actually scale.
Case Studies
B2B founders spend 4-5 hours a day on social media and outbound, battling API rate limits and IP bans.
Data Ingestion Hack
Bypassed expensive, restrictive platform APIs by using Serper API and Google site-search (e.g., site:reddit) to pull real-time competitor and trend data.
The Hermes Logic
A multi-agent engine with isolated decay schedules. Subagents run Deep Reddit Recon to extract native vocabulary, and Rival Spy to extract competitor weaknesses for "shadow positioning."
The Anti-Slop Editor
Hardcoded strict negative prompts stripping out corporate jargon (no "Delve" or "Unpack") and AI formatting.
The Shipping Layer
The engine queues drafts in Supabase. A custom Chrome Extension reads the queue and "ghost-types" the content directly into native social platforms using the user's IP, completely avoiding bans.
BachatMax
Structuring 110+ complex credit card reward rules, exclusions, and lounge limits into a usable consumer UI without an engineering budget.
Serverless ETL Pipeline
Built using n8n to ingest and normalize unstructured data into a 15-table Supabase relational database (tracking 100+ data points per card).
Proprietary Calculators
Built calculators in n8n (Spend Optimizer, Lounge Finder) that pull live Supabase data, run math functions, and push results to the UI.
Analytics Layer
Full analytics tracking implemented via PostHog.
Enterprise Voice AI Collections
Human collection agents are expensive, inconsistent, and pose massive RBI regulatory risks. Furthermore, CRM data gets stale because agents fail to log outcomes accurately.
4-Part n8n Orchestration
Built a master workflow combining a Calling Engine, Payment Link Engine (Razorpay), Confirmation Engine, and CRM Update Engine.
Regulatory Guardrails
Hardcoded logic to respect strict RBI call timing compliance (8am-7pm) and auto-reschedule missed calls.
The Voice Agent
Integrated Bolna AI to handle complex Hindi/Hinglish edge cases (Promise to Pay, rescheduling, human escalation) directly with borrowers.
Post-Call ETL
Used structured LLM extraction to instantly parse call summaries, outcomes, and next-call dates, pushing them directly back into the Google Sheets CRM.
View Strategy Deck
Loan Collection Voice Agent
Bolna + Paytm Integration
Manually creating SEO-optimized content at scale is time-intensive and inconsistent in quality.
Automated Scraping
Automated search engine scraping with constrained LLM generation for consistent, high-quality output.
CMS Integration
Pushed clean, CMS-ready content directly to Sanity via webhooks for seamless publishing.