I'm Ankit — an AI product builder with a 4+ year data engineering foundation. I turn ideas into working products: agentic AI platforms, automation, and native apps — from first prompt to production.
Hire me to design and build the AI product you need — or to add an intelligent layer to something you already run.
Multi-agent workflows that run themselves — research, writing, and analysis pipelines with parallel agents and real-time streaming. Automate the repetitive, human-in-the-loop where it matters.
From idea to a launch-ready product. I build full-stack AI apps with auth, payments, and a lean free-to-scale infra so you can validate fast.
Dashboards, pipelines, and reporting that actually get used. I've tracked $70M ARR and cut manual reporting work by 80%.
Add AI to your existing product — chat, summarization, extraction, RAG, structured outputs. Built to be reliable, not a demo.
Inbound & outbound voice agents that book, reschedule, and answer — deployed to production for a real dental clinic. Great for clinics, service businesses, and support.
Need to show investors or test an idea this week? I ship working, interactive prototypes fast — real UI, real AI, no mockups.
Flagship AI builds and shipped apps you can try right now.
Tracks your spending with zero manual entry. It reads bank SMS and UPI notifications on-device, detects every payment in real time, and pops a one-tap categorize dialog over any app the moment money leaves your account. No internet permission by design — financial data physically cannot leave the phone. A hand-tuned regex engine (trained on thousands of real Indian bank messages) extracts amount, merchant, and account; self-transfers and credit-card bill payments are auto-detected and excluded from totals; and categorizing one merchant retroactively tags every past payment from it. Try the interactive browser demo — it fires simulated bank notifications so you can experience the whole auto-detection flow without an Android phone.
Drop in a topic and watch a 7-phase agentic pipeline research it (parallel Tavily searches), outline it, write every section concurrently via map-reduce, score its own quality and rewrite the weak parts, auto-generate Mermaid charts from any numeric data, add AI cover images, and produce SEO — then publish to dev.to, Hashnode, LinkedIn, and Medium in one click. 50+ models across 7 providers behind a single dropdown with automatic fallback. Google OAuth, Fernet-encrypted API keys, and full draft history on Neon PostgreSQL.
Type any topic and four AI agents plan, generate queries, search the live web, and synthesise a structured, sourced report — every step streamed to the browser in real time. Built on a LangGraph graph with Send() fan-out for parallel section research, then converts the report to a blog post and publishes in one click.
AI-powered quiz and meme generation with a React frontend, Python backend, user auth, and real-time analytics. Live and usable today.
Meme generator with Google OAuth, drag-and-drop editing, comments, and leaderboards. Built with Flask and deployed to production.
A documentation/info site for the town of Gabhana with structured data, canonical URLs, and Open Graph tags — technical SEO done properly, deployed on GitHub Pages & Vercel.
Classic ML & automation projects — where the foundation was built
ML classification predicting APS failures with an ensemble of KNN, Random Forest, and XGBoost.
GitHub →Classification model for semiconductor wafer faults — a 24% productivity increase via ML-driven optimization.
GitHub →Voice-controlled personal assistant with NLP and task automation — years before "agentic AI" was a term.
GitHub →Selenium-based scraper that tracks product prices and deals automatically.
GitHub →Deep in AI and data, comfortable across the full stack.
I spent 4+ years in data — at Pluralsight, Tiger Analytics, and Infosys — building the dashboards and pipelines that $70M+ businesses run on. I know how to make data trustworthy, and how to turn it into decisions.
Now I build AI products. Not toy demos — working apps with real infrastructure, error handling, and the boring reliability details that separate a prototype from something a business can depend on. That's the same rigor I brought to analytics, pointed at a new stack.
For freelance clients that means one thing: you get a builder who ships, and who understands your data and your business — not just the AI hype.
B2B trial funnel analysis with Adobe Analytics & Snowflake; led a 5-phase Tableau hub audit and drove dashboard consolidation.
Built 30+ Tableau dashboards tracking $70M ARR, 622M impressions, and $9.4M ad spend. Automated reporting with Python + Snowflake, cutting manual work by 80%.
RFM analysis and uplift modeling for a retail client, enabling targeted promotions and sharper marketing efficiency.
Built CRM data pipelines and automated interface jobs in Python; improved operational efficiency by 30%.
Whether it's an AI SaaS MVP, a voice agent, or an intelligent layer on your existing product — tell me what you're trying to ship and I'll tell you how I'd build it.