Backend-first engineer and founder. I own messy, high-scale infrastructure problems end to end — and I build products on the side.

By day: data pipelines and AI tooling at Crustdata. Before that: carrier systems and telecom infrastructure at Plivo. Most of my highest-impact work lives in private repositories — this profile is intentionally outcome-led. Scale handled, systems shipped, problems solved are better proxies than commit volume. Private contributions are enabled.

Founding Nayikala — AI growth systems for Indian SMEs. And GigaCity — agrivoltaic platform turning farmland into shared clean energy and organic food infrastructure. Two different problems, same bias toward production systems over prototypes.

Building at the intersection of backend systems, data infrastructure, and AI-powered products. I care about craft: from backend reliability to the way a product actually feels when someone uses it.

• 📍 Bangalore, India · Open to remote
• 🎓 IIIT Hyderabad · JEE AIR 650 — top 0.05% of 1.2M candidates · INMO 2018 (top 50 nationally)

• Rebuilt a core enrichment pipeline: 50M → 200M+ rows/day (4×) — architecture, not brute-force scaling
• Automated carrier lifecycle: 1,200 → <100 support tickets/month, $300K/yr saved — systematic automation, not monitoring
• Shipped data products at scale: 1B+ people profiles, 60M+ companies, 654K+ parliamentary records
• Nayikala client deployments: -47% first-response time, -52% manual reporting, -38% support backlog

nayikala.com — AI growth systems for Indian SMEs (10–250 employees)

Indian businesses run on WhatsApp, spreadsheets, and phone calls. Nayikala audits the operational drag, then deploys AI automation systems around the way businesses already work — integrating WhatsApp, email, CRM, and accounting tools into unified workflows with human oversight and measurable KPIs.

Not consulting. Not POCs. Production systems that teams actually use daily.

Delivery model: Diagnose → Deploy → Improve. Fixed scope, outcome-disciplined. If numbers don't move, we keep working.

Client outcomes so far:

• -47% first-response time (WhatsApp lead unification)
• -52% manual reporting time (automated KPI summaries)
• -38% support backlog (AI-assisted task handling)
• +34% qualified inquiry quality (website rebuild + intelligent routing)

Flagship product: Rep4Real — demonstrates the full automation workflow: lead intake, WhatsApp integration, inquiry routing, follow-up automation. Built to be seen working, not just described.

Astro · Next.js · React · FastAPI · PostgreSQL · WhatsApp API · Resend · Vercel

India's farmland crisis and solar access gap are the same problem viewed from different angles. GigaCity puts solar panels 6–8 feet above farmland: electricity generated above, organic crops grown below. One piece of land, two income streams, three stakeholder groups.

The three problems it solves simultaneously:

• Farmers trapped in loan cycles, MSP anxiety, and middlemen — GigaCity gives them ₹3.21L/acre/year with buyback guarantees
• Urban dwellers and renters locked out of rooftop solar — GigaCity lets them buy stakes in real plants on real farmland
• Investors wanting real-asset exposure in renewables — 35–45% ROE, 1–2 year exit window, pre-sold demand before plants go live

One land. Clean energy. Organic food. Three stakeholders winning simultaneously.

Platform (Emprix) — full-stack system powering the entire GigaCity lifecycle:

• 6 user roles: consumer, gig worker, investor, contractor, private project owner, admin — each with bespoke onboarding state machines and permission layers
• Project lifecycle: SUBMITTED → APPROVED → LIVE → ICO_LIVE → COMPLETED with investor-exclusive funding windows, margin pool ROE calculations, and capacity snapshots
• Design workspace: AI-assisted collaboration with artifact versioning, step dependencies, comment threads — per-project
• Material marketplace: brand → model → technology → BOM costing (solar panels, inverters, batteries)
• Async infrastructure: Celery + Redis job queue, Alembic migrations, Docker Compose production stack
• 51 models · 23 services · 56 repositories · 160+ tests

Marketing & content: multi-page site (R3F 3D scroll, GSAP, Lenis), landing page (Framer Motion), Remotion promotional video, design system variations.

FastAPI · PostgreSQL · Redis · Celery · React 19 · TypeScript · Docker · Next.js · R3F · GSAP · Framer Motion · Remotion

Sansad QA v2 (active)

Parliamentary fact-checking: 654,539 Q&A records from Indian Parliament (1999–present). Paste a news article → decompose into atomic claims → retrieve evidence → SUPPORTED / CONTRADICTED / UNVERIFIABLE verdicts per claim. Fully local, no API keys.

What makes it interesting: the retrieval thinking — hybrid BM25 + vector search with Reciprocal Rank Fusion, domain-specific query expansion (parliamentary abbreviations, fiscal year normalization), and a proper eval framework (Recall@K, MRR, NDCG@K) to measure whether it actually works.

TypeScript · SQLite FTS5 · sqlite-vec · Ollama · Reciprocal Rank Fusion · Express ~6,500 lines · 6.9 GB corpus · <10ms hybrid search

Emprix (active — the platform powering GigaCity)

See the full technical breakdown under Founding → GigaCity above.

FastAPI · PostgreSQL · Redis · Celery · React 19 · TypeScript · Tailwind v4 · Docker · Alembic 51 models · 23 services · 56 repositories · 160+ tests

Island World Portfolio

Interactive 3D island world — entirely in the browser. Scroll-driven camera, Rapier physics, toon shaders, A* pathfinding, Web Audio beat detection, day/night cycle. Built because backend engineers should be able to ship things that are technically strong and actually pleasant to use.

Three.js · React Three Fiber · Rapier · GLSL · React 19

Court Piece Companion

Full-stack implementation of Court Piece (a South Asian trick-taking card game) with a Go game engine, AI bot opponent, and React PWA frontend. The interesting part: the bot implements the full game-state machine and rule engine, not just a lookup table. Go · React · TypeScript · Game AI

Social Content Studio

Single-brand content pipeline: brand calibration → Gemini image generation (3 variants per request, no-text/no-watermark) → caption + hashtag generation → export. Vitest suite covering unit, integration, and e2e-style flow tests. Next.js · TypeScript · Gemini API · Supabase · Vitest

Multi-Chat PDF System

Local PDF chat for family members who are government employees — sensitive documents that can't go to the cloud. Full-stack, runs entirely offline. Full-Stack · NLP · React · Local LLM

Linux Shell

Bash-like shell in C — piping, I/O redirection, process management, job control, command history, built-ins. C · Unix · Systems

• Nayikala — Scaling client delivery, deepening vertical playbooks, building toward productized SaaS
• GigaCity — Platform launch: stakeholder onboarding, live project dashboards, investor ICO mechanics
• Emprix — Design workspace, contractor management, private project funding flows
• Sansad QA — Public deployment: web app + browser extension for real-time parliamentary fact-checking
• AI Agent Architectures — MCP servers, multi-agent workflows, natural-language interfaces for data platforms

Languages

Backend & Data

Cloud & DevOps

AI / Creative

🐱 My GitHub Data

📦 1.2 MB Used in GitHub's Storage

🏆 1,099 Contributions in the Year 2026

🚫 Not Opted to Hire

📜 3 Public Repositories

🔑 26 Private Repositories

I'm a Night 🦉

🌞 Morning                319 commits         ██░░░░░░░░░░░░░░░░░░░░░░░   08.69 % 
🌆 Daytime                561 commits         ████░░░░░░░░░░░░░░░░░░░░░   15.29 % 
🌃 Evening                1488 commits        ██████████░░░░░░░░░░░░░░░   40.54 % 
🌙 Night                  1302 commits        █████████░░░░░░░░░░░░░░░░   35.48 % 

📅 I'm Most Productive on Wednesday

Monday                   693 commits         █████░░░░░░░░░░░░░░░░░░░░   18.88 % 
Tuesday                  650 commits         ████░░░░░░░░░░░░░░░░░░░░░   17.71 % 
Wednesday                699 commits         █████░░░░░░░░░░░░░░░░░░░░   19.05 % 
Thursday                 688 commits         █████░░░░░░░░░░░░░░░░░░░░   18.75 % 
Friday                   225 commits         ██░░░░░░░░░░░░░░░░░░░░░░░   06.13 % 
Saturday                 333 commits         ██░░░░░░░░░░░░░░░░░░░░░░░   09.07 % 
Sunday                   382 commits         ███░░░░░░░░░░░░░░░░░░░░░░   10.41 % 

📊 This Week I Spent My Time On

💬 Programming Languages: 
Markdown                 12 hrs 50 mins      █████████████░░░░░░░░░░░░   50.51 % 
Other                    3 hrs 31 mins       ███░░░░░░░░░░░░░░░░░░░░░░   13.85 % 
JavaScript               2 hrs 6 mins        ██░░░░░░░░░░░░░░░░░░░░░░░   08.26 % 
TypeScript               1 hr 38 mins        ██░░░░░░░░░░░░░░░░░░░░░░░   06.43 % 
HTML                     1 hr 36 mins        ██░░░░░░░░░░░░░░░░░░░░░░░   06.31 % 

🔥 Editors: 
Claude Code              25 hrs 24 mins      █████████████████████████   100.00 % 

💻 Operating System: 
Mac                      25 hrs 24 mins      █████████████████████████   100.00 % 

Last Updated on 08/06/2026 04:17:42 UTC