Software Engineer & Builder · IIIT Delhi (Computer Science & Design '28)
Building low-level systems, database architectures, autonomous AI engines, and production full-stack platforms.

I am a Computer Science & Design undergraduate at IIIT Delhi who builds across the entire software stack.

My work spans low-level systems and kernel drivers, database architectures and enterprise software, autonomous AI agent engines (MCP & LangGraph), and production full-stack SaaS platforms. I like tackling complex, real-world problems from scratch—whether that involves simulating packet drops inside the Linux kernel, architecting dual-database ACID systems, or building sub-second spatial engines.

Systems architecture case study simulating probabilistic packet drop inside the Linux Kernel 6.12 Intel E1000 NIC driver.

• Intercepted packet transmission at Layer 2/3 in struct sk_buff, silently freeing buffer memory via dev_kfree_skb_any() to trigger upstream TCP transport-layer recovery (Fast Retransmit, exponential RTO backoff).
• Benchmarked 100,000 transmitted packets across drop rates; proved that doubling host server receive buffers reduced recv() syscall invocations by ~50% due to stream-oriented socket coalescing (buf[64] vs buf[128]).
• Stack: C · Linux Kernel 6.12 · QEMU · Intel E1000 Driver · TCP/IP Networking
• Systems Case Study

Multiplayer board game engine and transactional backend built with Python, Flask, and MySQL 8.0.

• Designed an ACID-compliant relational schema tracking double-entry balance ledgers, dynamic rent multipliers, player-to-player property trade negotiations, and mortgage states.
• Implemented transactional concurrency controls and a RESTful API managing live game state for up to 6 players across 40 spaces.
• Built an interactive client with Vanilla JavaScript, server polling synchronization, and CSS variable theming.
• Stack: Python 3 · Flask · MySQL 8.0 · ACID Transactions · Relational DBMS · REST API
• GitHub Repository

Enterprise desktop ERP system built in Java Swing and Maven with dual-database transactional logic.

• Dual-Database Architecture: Engineered isolated authentication database (auth_schema.sql) and academic operations database (erp_schema.sql) to separate security boundaries from operational records.
• Built Role-Based Access Control (RBAC) supporting Student, Instructor, and Administrator roles with course registration/drop rules, section filters, grade auditing, and CSV exports.
• Stack: Java 17 · Maven · MySQL · Swing UI · RBAC · Dual-Database Architecture
• GitHub Repository

Cadastral GIS platform and automated statutory land verification engine for India's Dholera Special Investment Region (Tata Semiconductor corridor).

• Digitized and georeferenced 3,600+ survey numbers across 22 villages (18,161 cadastral parcels) with Town Planning demarcations and Satellite GIS overlays.
• Engineered a 1-click legal dossier compiler with SHA-256 provenance and instant QR verification (/verify/[id]), executing in-memory on the client to bypass serverless payload limits.
• Custom DPB1 delta-int32 binary serialization cutting payload size by >80% with off-thread Web Worker raycasting.
• Built a programmatic SEO architecture with 3,700+ Google-indexed parcel pages for organic zero-CAC discovery.
• Monetized with tiered recurring subscriptions (Razorpay auto-debit) and pay-per-report credits.
• Stack: Next.js 15 · TypeScript · MapLibre / Leaflet · pdf-lib · Supabase · Cloudflare Edge · Razorpay
• Live Platform · Architecture Case Study

Geospatial intelligence platform digitizing Delhi Development Authority (DDA) layouts and circle rates.

• Digitized 362+ official DDA town planning layout blueprints across all 36 Rohini sectors.
• 60 FPS imperative camera engine manipulating direct CSS matrix transforms with zero React render overhead.
• Built 705 SSG static routes pre-rendered via Next.js 16 Turbopack for instant sub-second page loads.
• Automated 2025 Delhi circle rate valuation engine and dynamic client-side PDF brochure generator.
• Stack: Next.js 16 · TypeScript · Turbopack · Leaflet · Supabase · Razorpay
• Live Platform · Architecture Case Study

Self-hosted lead scraping, multi-source contact enrichment, and cold outreach sequencer with native Model Context Protocol (MCP).

• Native MCP server allowing AI agents to autonomously trigger scraping, verification, and CRM sequencing.
• Stealth Google Maps scraping via Playwright with browser fingerprint rotation and anti-bot evasions.
• Multi-source contact enrichment (Apollo-style email permutation, DNS MX/SMTP server validation).
• Multi-mailbox cold outreach sequencer with Twenty CRM synchronization and 0–100 AI lead scoring.
• Stack: Python 3.11 · FastAPI · Playwright · MCP SDK · Docker · Next.js 15
• GitHub Repository

Local SEO crawler, site audit engine, and Answer Engine Optimization (AEO) scoring for Google AI Overviews and Perplexity.

• Native Model Context Protocol (MCP) agent server enabling LLMs to audit websites, inspect sitemaps, and trigger IndexNow submissions.
• Fast local web cockpit built with Hono, Cheerio, and React 19.
• Programmatic issue diagnosis: Core Web Vitals, canonical graph loops, and schema entity validation.
• Stack: TypeScript · Hono · MCP SDK · React 19 · Cheerio · IndexNow
• GitHub Repository

Local intelligent tutoring and textbook Q&A assistant using retrieval-augmented generation.

• Locally hosted language model inference via Ollama (Qwen 2.5:7B + BGE embeddings) with ChromaDB vector store.
• Multi-agent orchestration for grounded citation Q&A, automated quiz generation (MCQ/short answer), and misconception evaluation.
• Stack: Python 3.12 · Ollama · ChromaDB · Streamlit · RAG · Vector Embeddings
• GitHub Repository

A 6-agent LangGraph swarm analyzing multi-dimensional cultural affinity and sponsorship fit for brands and artists.

• Explores a 6-agent swarm (Entity Resolver, Taste Profiler, Friction Analyst, Mitigation Strategist, Compliance Auditor, Synthesis Agent).
• Vector affinity decomposition across 7 cultural domains using Qloo API with quantitative zone classification ($|Δ| < 0.30$ Green, $0.30 \le |Δ| < 0.60$ Yellow, $\ge 0.60$ Red).
• Built with strict 0-PII compliance and contract-tested with Pytest.
• Stack: Python 3.12 · LangGraph · FastAPI · Qloo API · React 19 · Pydantic v2
• GitHub Repository

Growth engine built by reverse-engineering Twitter's Heavy Ranker recommendation algorithm and X-AI ranking mechanics.

• 7-point pre-flight anti-shadowban linter evaluating engagement multipliers and penalty weights before publishing.
• Zero-touch repository and domain ingestion to turn code updates into engaging technical narratives.
• Jitter scheduling to mimic human engagement patterns and protect sender reputation.
• Stack: React 19 · TypeScript · Tailwind CSS · Vite · Algorithms
• GitHub Repository

Autonomous ATS crawler and job discovery pipeline with programmatic single-page resume compilation.

• Automated ATS fetchers streaming live postings from Greenhouse, Lever, and Ashby JSON endpoints.
• Deterministic Gate 1 filter (<1ms) and SQLite caching for rapid opportunity triage.
• Programmatic Typst / RenderCV resume compiler with automated single-page length enforcement test gates.
• Trigger-event personalized outreach draft generator.
• Stack: Python 3.11 · Typer CLI · SQLite · Typst · RenderCV · Pytest
• GitHub Repository

• Systems & Languages: C, Linux Kernel (Device Drivers, sk_buff, Sockets), Java 17, Python 3.12, TypeScript, SQL (MySQL 8.0, PostgreSQL, SQLite), Typst
• Databases & Architecture: ACID Transactions, Normalized Relational Modeling, Dual-Database Isolation, Supabase, Redis, Connection Pooling
• AI Agents & Protocols: Model Context Protocol (MCP Native), LangGraph, LangChain, ChromaDB, Ollama, Vector Embeddings, Playwright Automation
• Frontend & Geospatial: Next.js (App Router), React 19, MapLibre GL, Leaflet, Web Workers, Turbopack, Tailwind CSS v4, Framer Motion
• Backend & Cloud: FastAPI, Node.js, Express, Flask, Pydantic v2, RESTful APIs, Cloudflare (Edge/Cache), Docker, Vercel

• B.Tech in Computer Science and Design (CSD) — IIIT Delhi (2024 – Present)
• Organizing & Leadership:
  • Odyssey IIITD: Managed tech registration workflows, server polling systems, and on-ground logistics for 10,000+ attendees.
  • E-Summit IIITD: Organized startup pitch cohorts, founder panels, and jury evaluation workflows.

• Portfolio: www.webforge.me
• LinkedIn: linkedin.com/in/aryan-bansal-b29b69370
• Email: aryan24120@iiitd.ac.in · aryanbanc@gmail.com

_{© 2026 Aryan Bansal · Designed with care}