SELECTED ENGINEERING WORK

ENGINEERED SOFTWARE & AGENTIC SYSTEMS

I turn operational problems into software that can ship, scale, and keep working—from full-stack web architectures and custom Micro-SaaS backends to autonomous AI agents and automated workflow pipelines.

Each case below shows the problem, the system built to solve it, the engineering decisions behind it, and the honest stage of the work.

AI-powered career platform · Live

SkillForge

Bridging the Gap Between CS Education and Industry Readiness.

THE PROBLEM

Computer science learners often lack a clear, evidence-based path from academic knowledge to industry-ready skills.

THE SOLUTION

A six-domain assessment and roadmap platform with data-driven gap analysis and a retrieval-augmented learning assistant.

FEATURES & DELIVERABLES

Multi-domain technical skill assessments
Data-driven skill gap analysis
Personalized milestone-based roadmaps
RAG-powered AI learning assistant

SkillForge AI career development platform banner

SYSTEM ARCHITECTURE

Next.js and TypeScript interface connected to a containerized FastAPI and Python backend, Supabase persistence, and LangChain-powered retrieval workflows.

AI SEEKHO capstone · Live

AI Startup Launcher

An autonomous AI board of directors analyzing your risk and revenue strategy.

THE PROBLEM

Startup validation requires coordinated research, product, pricing, and go-to-market decisions that are often handled in disconnected workflows.

THE SOLUTION

A supervised multi-agent platform that delegates specialist work, preserves shared state, and keeps strategic decisions under human review.

FEATURES & DELIVERABLES

Market Research Agent
Product Strategist Agent
Pricing Agent
Marketing Agent
Orchestrator Agent

AI Startup Launcher multi-agent platform banner

SYSTEM ARCHITECTURE

Next.js client connected to a Python FastAPI service, LangGraph orchestration, LangChain tools, Groq inference, and Supabase persistence.

Emergency response platform · Live

ErythroNet

Emergency Blood Verified Donors, Direct to Hospital.

THE PROBLEM

Panic broadcasts are slow, difficult to verify, and poorly targeted when a patient urgently needs blood.

THE SOLUTION

A real-time donor matching and emergency request system with location-aware alerts, resilient delivery paths, and hardened API controls.

FEATURES & DELIVERABLES

Real-Time Matchmaking
Live Emergency Board
Instant Area Alerts (PWA/FCM)
Resilient Architecture with caching & fallback logic
Strict security with Rate Limiting, Helmet, and CORS

ErythroNet emergency blood donation network banner

SYSTEM ARCHITECTURE

React and TypeScript PWA connected to a Node.js and Express API, Firebase Firestore, Firebase Cloud Messaging, and resilient cache and fallback layers.

Tool-augmented AI agent · Live demo

Cheezious AI Agent

Re-imagining QSR Customer Experience with Autonomous AI Agents.

THE PROBLEM

QSR customers need fast, contextual answers across branches, menus, budgets, billing, and complaint workflows.

THE SOLUTION

A ReAct-based assistant that selects purpose-built tools for live search, location discovery, menu analysis, bill calculation, and complaint dispatch.

FEATURES & DELIVERABLES

Real-Time Web Search (SerpAPI)
Google Maps Branch Locator
Menu Category Analyzer
Budget Deal Recommender
GST Tax & Bill Estimator
Automated Complaint Logging & Email Dispatch via yagmail

Cheezious autonomous QSR AI agent banner

SYSTEM ARCHITECTURE

Streamlit interface connected to a LangGraph and LangChain ReAct workflow, Groq inference, SerpAPI search, mapping tools, and yagmail dispatch.

VIEW ALL WORK ON GITHUB ↗