Available for internships

Creating Intelligent
Web Experiences

Computer Science student building thoughtful digital experiences with AI-assisted development and modern web technologies.

LeetCodeLet's Connect

Background

I am a curious developer and Computer Science student learning to build useful software.

Currently in my second year at SNS College of Technology, my focus is on practical problem-solving. I am deeply interested in modern web development and UI/UX design, and I learn best by building real-world projects rather than just reading documentation.

I am a strong advocate for AI-assisted development. I actively use tools like Claude, ChatGPT, and Gemini to generate code, which accelerates my ability to prototype ideas. My actual work lies in understanding that code, testing it, modifying the business logic, and deploying it to production.

I don't claim to know everything yet, but I learn incredibly fast. My goal is to continually improve my understanding of systems architecture, ask the right questions, and eventually contribute to products that genuinely impact how people work and live.

University

B.E Computer Science & Engineering

SNS College of Technology

Second Year • 2024–Present

School

Vidya Vikasini Matric Higher Secondary School

Higher Secondary Education

Core Stack

Next.js & React

TypeScript

Python

HTML / CSS

Design & Tools

UI/UX Design

FlutterFlow

Canva

GitHub

Languages

English

Tamil

Malayalam

Currently BuildingActive Prototype

AgentLens

Autonomous AI agent observability platform providing execution graph visualization, LLM call tracing, and real-time failure diagnostics.

Technical Exposure

Skills &
Workflow

I build modern web applications by combining solid engineering fundamentals with AI-assisted development, rapid prototyping, and iterative refinement. Every project emphasizes maintainability, performance, and user experience.

Frontend

ReactNext.jsTailwind CSSFramer Motion

Backend

FastAPISupabasePostgreSQLREST APIs

Deployment

Cloudflare PagesGitHub ActionsGoDaddy DNS

AI & Workflow

ClaudeGeminiChatGPTAntigravity

Selected Work

Featured Projects

AgentLens Preview
01

AgentLens

AI Observability & Trace Debugging

Jump
Artist Color Lab Preview
02

Artist Color Lab

Computational Color & Studio Tools

Jump
Aerova Preview
03

Aerova

Macroeconomic Data & Time-Series

Jump
Automotive Website Sample Preview
01Sample

Automotive Website Sample

Sample Website

Jump
Cafe Website Sample Preview
02Sample

Cafe Website Sample

Sample Website

Jump
Gym Website Sample Preview
03Sample

Gym Website Sample

Sample Website

Jump
Category 1

REAL PROJECTS

Software engineering projects, academic research prototypes, and production systems where I led or contributed directly to technical architecture and implementation.

01 • Academic / Mini ProjectAI Observability & Trace DebuggingWorking Prototype
GitHub

AgentLens

Autonomous AI agent observability platform providing execution graph visualization, LLM call tracing, and real-time failure diagnostics.

AgentLens • Academic / Mini Project
Real UI Demo
Click to play
Real UI capture showing prompt interception, autonomous tool calls, and node execution latency.Actual UI Flow
Interactive Screen RecordingAspect Ratio: 16:10 • Audio Enabled
Role & Contribution

System architecture, drop-in SDK instrumentation, and interactive execution graph visualization.

The Problem

Autonomous AI agents execute multi-step reasoning loops and external tool calls in the background. When an agent enters an infinite loop, hallucinates, or fails a tool execution, developers lack clear step-by-step diagnostic telemetry to understand what went wrong.

Approach & Implementation

Designed a visual debugging workbench for AI agents. Built an interactive decision-tree graph that deconstructs agent reasoning chains into discrete nodes with payload metrics and a two-line drop-in SDK integration.

Key Features & UX Flow
3 Core Modules
Feature 01
Visual Execution Graph

Maps autonomous agent decisions into an interactive node flow with latency and payload metrics.

Feature 02
Drop-in SDK Instrumentation

Two-line initialization capturing prompts, tool calls, and state mutations across major LLM providers.

Feature 03
Live Observability Console

Real-time session stream to inspect execution traces and diagnose tool timeouts.

Observed Workflow Steps
1User prompt initiates execution trace with model parameter telemetry
2Graph visualizes decision-tree node branching into external tool calls
3Console flags tool timeouts and payload exceptions in real time
Technologies & Stack
Next.jsTypeScriptTailwind CSSFramer MotionOpenAI SDKAnthropic SDK
Key Architectural Takeaway

Architected asynchronous streaming graph state to visualize multi-step agent reasoning without UI stutter.

GitHub
02 • Internship Project • Arkensys RealtorsComputational Color & Studio ToolsLive Application
Live DemoGitHub

Artist Color Lab

Precision paint mixing calculator and custom palette studio translating digital color values into physical acrylic recipes.

Artist Color Lab • Internship Project
Real UI Demo
Click to play
Real UI capture showing physical acrylic tube selection, passwordless studio auth, and 5-step mixing recipes.Actual UI Flow
Interactive Screen RecordingAspect Ratio: 16:10 • Audio Enabled
Role & Contribution

Software engineering intern at Arkensys Realtors; developed subtractive pigment mixing algorithms, studio palette management, and passwordless authentication.

The Problem

Digital screen colors do not map directly to real-world paint pigments. Artists waste expensive acrylic tubes through trial-and-error mixing to match a reference using whatever specific paint tubes they physically have on hand.

Approach & Implementation

Built a parts-based calculation engine that computes exact proportional mixing ratios based on the artist's real inventory, coupled with passwordless studio storage to persist custom palettes across devices.

Key Features & UX Flow
3 Core Modules
Feature 01
Parts-Based Mixing Engine

Calculates practical mixing ratios based on the physical acrylic tubes in the artist's inventory.

Feature 02
Custom Palette Studio

Create, save, and organize personalized pigment sets for ongoing studio projects.

Feature 03
Passwordless Authentication

Email OTP authentication ensuring palettes persist seamlessly across desktop and mobile.

Observed Workflow Steps
1Select acrylic tube brands physically owned in studio inventory
2Pick reference colors on screen or upload source reference photos
3Generates clear parts-based recipes (e.g. 3 parts Ultramarine, 1 part Titanium White)
Technologies & Stack
Next.jsReactTailwind CSSSupabaseColor Mixing Algorithms
Key Architectural Takeaway

Engineered subtractive pigment approximation algorithms and responsive state management for tactile studio use.

Live ApplicationGitHub
03 • SIH 2026 / Team ProjectMacroeconomic Data & Time-SeriesWorking Prototype
GitHub

Aerova

Real-time domestic airfare index and market intelligence platform augmenting Consumer Price Index (CPI) calculations with live booking-window analytics.

Aerova • SIH 2026 / Team Project
Real UI Demo
Click to play
Real UI capture showing national airfare trends, T+1 to T+45 booking horizons, and DGCA passenger weighting.Actual UI Flow
Interactive Screen RecordingAspect Ratio: 16:10 • Silent UI Capture
Role & Contribution

Team frontend and data visualization lead for Smart India Hackathon 2026; built national route indexing, advance purchase booking windows (T+1 to T+45), and DGCA route weighting visualizations.

The Problem

Dynamic airline yield management causes rapid fare fluctuations that official monthly CPI inflation indices fail to reflect in real time, leaving travel managers and economic researchers without high-frequency price transparency.

Approach & Implementation

Engineered a real-time airfare index platform tracking national route price evolution. Structured advance-purchase booking horizons (T+1 to T+45) and weighted market indices using DGCA passenger traffic volume data.

Key Features & UX Flow
3 Core Modules
Feature 01
AERIX National Index

Baseline price series tracking price shifts with 7-day trend history and live telemetry.

Feature 02
Advance Purchase Windows

Tracks pricing volatility across T+1, T+7, T+15, T+30, and T+45 booking horizons.

Feature 03
DGCA Route Weighting

Top-volume trunk routes (DEL-BOM) normalized against Directorate General of Civil Aviation passenger traffic.

Observed Workflow Steps
1Live index hero metric with 7-day historical trend spline charts
2Granular fare decomposition separating base fares, taxes, UDF, and surcharges
3Multi-tab view covering Route Intelligence, Lead-Time Analysis, and data quality
Technologies & Stack
Next.jsTypeScriptTailwind CSSTime-Series VisualizationStatistical Indexing
Key Architectural Takeaway

Deconstructed multi-layered airline pricing models into clean, executive financial visualizations.

GitHub

Engineering Evidence

Behind the Build

Real architectural decisions, development evidence, and algorithmic solutions behind the projects.

AgentLens • Real UI Trace
Verified UI Flow
Click to play
Tested & Verified UI ImplementationArchitecture Evidence
Architecture EvidenceAI Telemetry & Execution Tracing

Visualizing Autonomous Agent Decision Loops Without Telemetry Lag

When building AgentLens, the core engineering hurdle was streaming asynchronous reasoning steps and LLM provider payloads into an interactive decision tree without blocking the frontend render loop. Built a lightweight drop-in SDK that intercepts model prompts and tool calls in real time.

Technical Architecture & Decisions

1
Decision Tree State Streaming

Buffers incoming node telemetry in memory and renders interactive directed graphs with per-node latency benchmarks.

2
Two-Line SDK Instrumentation

Wraps OpenAI and Anthropic SDK callers to automatically extract token counts, parameters, and tool call payload exceptions.

3
Real-Time Failure Diagnostics

Flags infinite loops and external API timeouts directly in the visual workbench console.

Next.jsTypeScriptTailwind CSSFramer MotionOpenAI SDKAnthropic SDK
GitHub

Career & Validation Archive

Certificates & Experience

Proof of the work, learning, and experiences behind the portfolio.

EXPERIENCE

Arkensys Realtors

Software Developer Internship

June 18–30, 2026

CERTIFICATIONS

Python & Programming

Software Engineering & Algorithms

Scaler Academy

Design Thinking & UX

Enterprise Practitioner & Workshop

IBM Enterprise DT

Hackathons & Competitions

Symposiums & Project Expos

HackNext • Technovibe • Oblivion

Honors & Athletics

Leadership & Athletic Excellence

Zone Sports Council

Every verified certificate, offer letter, and engineering document available in-page

Engineering Mindset

Engineering Perspective

How building, debugging, experimenting, and solving problems have shaped the way I think about software.

CORE WORKING PRINCIPLES
[06 PERSPECTIVES]
01
Problem Space & Constraints

Think in Problems, Not Just Technologies.

A technology is only useful when it helps solve a real problem. I am learning to understand the problem, constraints, inputs, outputs, and trade-offs before deciding how to build the solution.

02
Computational Thinking

Algorithms Shape How Software Thinks.

Learning data structures, algorithms, and computational thinking has changed how I approach problems. The goal is not just to make something work, but to understand why a solution works and how efficiently it scales.

03
Systems & Lifecycle

Software Is More Than Code.

A working program is only one part of an engineering solution. Architecture, data flow, error handling, testing, deployment, maintainability, and user experience all influence whether software actually works in the real world.

04
AI & Verification

AI Accelerates Development — Understanding Remains Essential.

Tools such as Claude, ChatGPT, and Gemini help me explore ideas and accelerate implementation. But generated code still needs to be understood, tested, modified, debugged, and integrated responsibly.

05
Diagnostic Method

Debugging Is Part of Thinking.

Some of the most useful learning happens when something does not work. Tracing an error, isolating the cause, testing assumptions, and iterating toward a solution has become an important part of how I learn.

06
Interdisciplinary Breadth

Computer Science Connects Many Disciplines.

Software development brings together algorithms, systems, mathematics, human interaction, data, and design. I want to keep exploring those connections rather than limiting myself to a single development domain.

Beyond Code

The human side outside the editor
♟

Chess

Calculation, strategic patience, and tactical discipline.

📖

Reading

Books on system architecture, design philosophy, and sci-fi.

✏

Drawing

Pen sketches, perspective studies, and visual composition.

Let's build something great.

I'm always looking for ambitious projects and exciting opportunities. Let's start a conversation or explore my professional presence.

Send a Message

Usually replies within 24 hours.

GitHub

Open-source projects, experiments and code.

LinkedIn

Experience, education and achievements.

LeetCode

Data Structures, Algorithms and practice.

Instagram

Personal updates and moments.

Email

prabhavkrishna17@gmail.com

Location

Coimbatore, Tamil Nadu, India

Interactive Speed Challenge

Think you can type faster than me?

Take the 30-second challenge and see where you rank.

30-Second Sprint

Enter your name to begin

Type real software engineering, system design, and AI observability sentences from my portfolio projects. The 30-second timer begins only after you click Start.

Sample Passage Preview

"Architecting autonomous AI observability platforms requires intercepting model prompts, instrumenting background tool calls, and visualizing decision tree flow graphs in real time without introducing UI latency."

Global Leaderboard

Top 10 Scores

No scores recorded yet.

Be the first participant by completing the 30-second sprint!

Real-time persistence powered by Supabase