SS· Resume
Resume

Developer Portfolio

Shashank
Sekhar VS

AI Engineer & Full Stack Developer

I build AI-powered applications and web systems from agentic workflows and RAG systems to full-stack products.

View selected work Get in touch →
Shashank Sekhar VS

01 — About

The Roots

Computer Science graduate working across three disciplines that are usually kept apart: applied AI, production web engineering, and UX research. Currently freelancing as a web developer while leading both the technical and design tracks of my college's department club.

The Approach

My work leans toward systems that don't just predict but act, combining forecasting models with reinforcement-learning agents that make operational decisions. My web work favours Next.js and Supabase, built for real tenants and real traffic, not demos. My design work starts from primary research: surveys, interviews, and usability testing, before a single screen gets drawn.

The Promise

But more than any one discipline, what I love is building: taking a rough idea and pushing it all the way to something real that people can open, click through, and actually use. That moment when a feature finally works exactly as imagined never gets old, and it's the reason I keep choosing projects that ship rather than projects that stay theoretical.

02 — Experience

Where I've worked

  1. Jan 2026 – May 2026

    Freelance Web Developer

    Built a full-scale webpage end-to-end and handled its SEO.

  2. Jan 2025 – Dec 2025

    Technical Lead — Department Association

    Led the technical team across club projects: code quality, system design, and mentoring junior developers in frontend and best practices.

03 — Selected Work

Things I've shipped

01

FlowForge — Autonomous Task Decomposition & Execution Engine

PERSONAL · AI ENGINEERING
  • Six-stage agent pipeline that analyzes a task, plans implementation, executes it, tests the result, and independently reviews the final artifact.
  • Role-specialized agents (plan, research, draft, review) hand off work with a built-in feedback loop.
  • Built with CrewAI and the Anthropic API, with quality-threshold exit conditions.
PythonAnthropic APIMulti-agentAgentRunner
View project →
02

Hybrid Framework for Predictive & Adaptive Supply Chain Management

ACADEMIC · AI/ML
  • Hybrid system: three CatBoost models forecast sales, delivery time, and profit per order.
  • Those forecasts feed a PPO reinforcement-learning agent, implemented from scratch in NumPy, that decides reorder quantities.
  • Evaluated on 180,519 records from the DataCo Smart Supply Chain dataset: mean reward 1843.0 vs. 1342.7 for a fixed reorder-point baseline.
CatBoostPPO / RLPythonTime-series
View project →
01

Cycle World Pondicherry — Headless Cycle Storefront

CLIENT · PRODUCTION WEB
  • Built and deployed a production e-commerce storefront for a bike dealership using Next.js, Supabase and PostgreSQL, with SSR, SEO-focused routing and database-level access control.
  • Achieved sub-second TTFB on catalogue pages and handled the project from implementation through deployment.
Next.jsSupabasePostgres RLSVercel Edge
View project →
02

QLess — Campus Pre-Order Web App

PERSONAL · WEB DEVELOPMENT
  • The pre-order and pickup flow from the QLess UX case study, rebuilt as a working web app.
  • Menu browsing, cart, and slot booking against real capacity limits.
  • OTP pickup confirmation with a live countdown, one HTML file, no framework, no backend.
HTML5CSS3Vanilla JavaScriptState management
View project →
01

Frabe — Women's Personal Safety App

UX/UI · CASE STUDY
  • Driven by primary research: 80 survey respondents and 12 in-depth interviews.
  • Solved reaching emergency help in under 8 seconds via a one-tap SOS flow and a safe-area map covering 9+ location categories.
  • High-fidelity Figma screens, stress-tested through Maze usability testing.
User researchFigmaUsability testing
View case study →
02

QLess — Campus Pre-Order & Queue Management App

UX/UI · CASE STUDY
  • Mobile pre-ordering and live token tracking that removes physical queues at high-footfall campus dining venues.
  • Full journey mapping from browse to collection, grounded in research with canteen visitors.
  • 10 high-fidelity screens built around a sub-3-tap ordering principle.
User researchJourney mappingFigma
View case study →

04 — Known Technologies

A connected stack

These technologies work together to help me build AI-powered and full-stack applications.

Build
Learn
Iterate
🧠 AI / ML
LangChain
LangChain LangChain

Framework for building LLM-powered agents with tool-calling and memory.

Key areas

  • Tool-calling integrations
  • Agent orchestration
  • Persistent memory
Anthropic API
Anthropic Anthropic API

The model layer behind every agent I've shipped, Claude via direct API calls.

Used in

FlowForge

Key areas

  • Agent reasoning
  • Structured output
  • Multi-agent coordination
Python
Python Python

My primary language for AI/ML work: modeling, agents, and data pipelines.

Used in

FlowForgeHybrid Framework

Key areas

  • ML modeling (CatBoost, PPO)
  • Agent pipelines
  • Data processing
OpenAI
OpenAI OpenAI

Familiar with the OpenAI API and ecosystem for LLM-based development.

Key areas

  • Chat completions
  • Embeddings
  • Function calling
🖥️ Frontend
Next.js
Next.js Next.js

React framework used for server-rendered, SEO-optimized product pages.

Used in

Cycle World

Key areas

  • Server-side rendering
  • Dynamic routing
  • Edge deployment
React
React React

Component-based UI library for building interactive frontends.

Key areas

  • Component architecture
  • State management
  • Hooks
JavaScript
JavaScript JavaScript

The language behind every client-side interaction I build, framework or not.

Used in

QLess Platform

Key areas

  • DOM manipulation
  • State without a framework
  • Async / event handling
Tailwind CSS
Tailwind CSS Tailwind CSS

Utility-first CSS for fast, consistent styling.

Key areas

  • Utility-first styling
  • Responsive design
  • Design tokens
🗄️ Backend
Supabase
Supabase Supabase

Postgres-backed backend-as-a-service: auth, database, and realtime.

Used in

Cycle World

Key areas

  • Row-Level Security
  • Realtime sync
  • Auth
PostgreSQL
PostgreSQL PostgreSQL

Relational database for structured, relationship-heavy data.

Used in

Cycle World

Key areas

  • Schema design
  • Row-Level Security
  • Query optimization
Node.js
Node.js Node.js

JavaScript runtime for backend services and tooling.

Key areas

  • REST APIs
  • Server-side logic
  • Tooling scripts
FastAPI
FastAPI FastAPI

Python web framework for fast, typed backend APIs.

Key areas

  • API design
  • Async endpoints
  • Data validation
☁️ Tools / Infra
Git & GitHub
Git Git & GitHub

Version control and hosting for every project I ship, including this site.

Used in

Every project

Key areas

  • Version control
  • Collaboration
  • CI-friendly workflows
Vercel
Vercel Vercel

Deployment platform for edge-cached, production web apps.

Used in

Cycle World

Key areas

  • Edge caching
  • CI/CD deployment
  • Preview environments
Figma
Figma Figma

Primary tool for UX/UI design, wireframes through high-fidelity screens.

Used in

FrabeQLess Design

Key areas

  • Wireframing
  • Prototyping
  • Design systems
Docker
Docker Docker

Containerization for consistent, portable environments.

Key areas

  • Containerization
  • Environment consistency
  • Local dev parity
🎓 Also familiar with Java · C++ · C · Dart · HTML5 · CSS3 · Flutter · Flask · MongoDB · MySQL · SQLite · Firebase · Kubernetes · Hugging Face

05 — Education

Where I've studied

  1. 2022 – 2026

    Bachelor of Technology

    Sri Manakula Vinayagar Engineering College

06 — Contact

Let's talk.

Open to AI Engineering and Full-Stack Developer opportunities.

Email Email LinkedIn LinkedIn GitHub GitHub Phone