// B.TECH COMPUTER SCIENCE • INFORMATION SECURITY • VIT VELLORE

Building software at the intersection of AI, Security, and Full-Stack Engineering.

I am a Computer Science student at Vellore Institute of Technology (VIT) specializing in Information Security. I build software systems across multi-agent orchestration, cyber threat intelligence, and multimodal dataset engineering — engineered with architectural clarity, modularity, and restraint.

Explore Flagship Projects Download Resume

ACADEMIC MERIT 8.27 CGPA VIT Vellore (2024–2028)

INDUSTRY EXPOSURE 10X Tech AI/ML & SWE Intern

SYSTEMS BUILT 4 Systems Flagship Codebases

CORE FOCUS InfoSec & AI Threat Intel & Agentic

Pranay Kumar Vonamala Computer Science Engineering • 3rd Year

Hyderabad • Vellore

CHAPTER 01 // HOW I APPROACH SOFTWARE

Craftsmanship, discipline, and understanding the system.

I view software engineering not as stringing packages together, but as constructing deterministic, reliable machinery. My background in Information Security establishes a zero-trust mindset toward every layer: from data schemas and cryptographic verification to concurrency control and memory persistence.

Outside of software, I maintain a consistent athletic and strength training discipline. The patience, progressive consistency, and technical focus required in the gym directly reinforce how I debug complex systems, write clean abstractions, and ship maintainable code.

Solve the real problem first

Isolate the true bottleneck or failure mode before writing code. An architectural rethink is almost always cleaner than adding defensive patches to flawed assumptions.

Modularity & deterministic state

Tightly coupled systems break silently. Isolate responsibilities with strict interfaces, clear data contracts, and reproducible states so components can be validated independently.

System fundamentals over hype

Libraries and tools evolve rapidly; foundational computer science remains constant. Deeply understand operating system primitives, database indexing, and network protocols.

AI as disciplined instrumentation

Language models belong inside deterministic engineering control planes — guarded by validation gates, schema enforcement, and structured retrieval — not unconstrained loops.

CHAPTER 02 // PROFESSIONAL WORK

Engineering Experience

10X TECHNOLOGIES

AI/ML & Full-Stack Developer Intern

August 2026 – October 2026

Contributed to AI/ML and software engineering initiatives involving production AI/ML models, Large Language Models (LLMs), and Retrieval-Augmented Generation (RAG) systems. Built and integrated application programming interfaces while contributing across frontend, backend, and UI/UX engineering.

•

AI & RAG Systems Integration: Developed AI-driven application workflows utilizing Hugging Face transformer models and LangChain for structured retrieval and semantic parsing.

•

Backend & API Development: Engineered robust REST API microservices, request validation pipelines, and data models to connect client interfaces with inference workloads.

•

UI/UX & Frontend Engineering: Built responsive user interface components, interactive data visualizations, and telemetry views ensuring low latency and clean state management.

•

Quality & Testing Rigor: Collaborated on model output evaluation, component regression testing, and code quality reviews to maintain secure data handling across environments.

Hugging Face LangChain Large Language Models RAG Systems REST APIs FastAPI React UI/UX

CHAPTER 03 // FLAGSHIP SYSTEMS

Engineering Projects

Four distinct software systems engineered from first principles to solve concrete problems across personal finance, cybersecurity, agentic workflows, and dataset generation.

SYSTEM 01 AI PERSONAL FINANCE PLATFORM

Repository Live Application

Spendly

AI-powered personal finance platform providing real-time expense tracking, automated budget allocation, predictive cash-flow forecasting, and conversational financial assistance.

THE PROBLEM

Manual logging apps feel tedious and fragmented, forcing users to input raw numbers without gaining forward-looking budgeting velocity or actionable guidance.

ENGINEERING SOLUTION

Engineered a reactive React SPA backed by Firebase Firestore real-time sync and Python FastAPI microservices. Integrated OpenRouter LLMs with custom system prompts to calculate savings velocity and provide contextual advice.

React FastAPI Firebase Auth Firestore OpenRouter API Recharts Framer Motion

SYSTEM ARCHITECTURE

React Client SPA Real-time UI • Recharts charts

Firebase Core User Auth • Firestore NoSQL

FastAPI Service Asynchronous Analytics Worker

04
                      OpenRouter LLM
                      Contextual Advisory Engine
                    

Actionable Dashboard Predictive Spending Forecasts

SYSTEM 02 CYBER THREAT INTELLIGENCE (CTI) PLATFORM

Repository

Sentinel AI

Cyber Threat Intelligence platform engineered to detect, correlate, and explain digital threats across raw indicators, suspicious URLs, domains, screenshots, and OSINT telemetry feeds.

THE PROBLEM

Security analysts suffer from alert fatigue and disjointed OSINT feeds, spending hours manually cross-referencing WHOIS records and reputation databases without unified context.

ENGINEERING SOLUTION

Built an asynchronous FastAPI pipeline integrating VirusTotal, WHOIS, and domain reputation feeds in parallel. Integrated an Explainable AI module that correlates indicators into structured risk scores with human-readable rationale.

Python FastAPI VirusTotal API OSINT Recon IOC Management Explainable AI Threat Telemetry

TELEMETRY PIPELINE

Threat Ingestion URL / IP / Domain / Media

OSINT Telemetry VirusTotal • WHOIS APIs

03
                      Explainable AI
                      Feature Correlation Engine
                    

IOC Hash Store Threat Correlation Scoring

SOC Console Plain-Language Threat Rationale

SYSTEM 03 AGENTIC WORKFLOW & RAG SYSTEM

Repository

SupportFlow AI

Customer support automation platform built with LangGraph state graphs, RAG, SQLite conversational persistence, and specialized agent routing using local open-weight LLMs.

THE PROBLEM

Single-prompt chatbot solutions fail to handle non-linear multi-step tickets, hallucinate company policy, and risk exposing sensitive internal customer data to public cloud APIs.

ENGINEERING SOLUTION

Engineered a stateful agent graph with conditional routing across dedicated sub-agents (Sales, Technical, Billing, Account). Integrated local RAG retrieval with Ollama runtime execution to ensure complete data privacy.

Python LangGraph LangChain RAG SQLite State Store Ollama Conditional Routing

AGENTIC STATE GRAPH

Customer Query Inbound Multichannel Ticket

Intent Supervisor Conditional Decision Router

03
                      Specialist Agents
                      Sales / Tech / Billing Nodes
                    

RAG • SQLite State Context Retrieval & Checkpoint Memory

Verified Resolution Local LLM Output (Zero Cloud Egress)

SYSTEM 04 VIDEO-TO-TEXT & DATASET PLATFORM

Repository

Textora Engine

Video-to-Text and multimodal dataset engineering platform designed to turn raw video sources into clean, validated, reproducible, and provenance-aware AI datasets.

THE PROBLEM

Raw video transcription suffers from rolling caption flicker, non-lexical audio tags ([Music], [Applause]), language drift (Hinglish/transliterated Latin scripts), redundant Whisper compute, and missing data lineage.

ENGINEERING SOLUTION

Engineered a 4-layer platform featuring caption-first dual ingestion (Whisper fallback), deterministic n-gram de-flickering, script validation gates, xxHash/SHA-256 deduplication, SQLite WAL queues, and multimodal temporal frame alignment.

Python 3.12 FastAPI faster-whisper SQLite WAL Queue xxHash64 • SHA-256 RAG Chunker Lineage DAGs

DATASET PIPELINE

Source Discovery YouTube • Playlists • Local

Caption-First Ingestion faster-whisper Fallback Engine

03
                      Normalizer & Gates
                      De-flicker • Script Ratio Validation
                    

Multimodal Grounding Temporally Aligned Frames

RAG Payloads SHA-256 Lineage Manifests

CHAPTER 04 // CAPABILITY ARCHITECTURE

Interactive Engineering Map

Click any technology to inspect what it is, how Pranay applied it, and the exact software system it powers.

AI & AGENTIC SYSTEMS

CYBERSECURITY & THREAT INTELLIGENCE

SYSTEMS, DATA & BACKEND

CORE LANGUAGES & WEB

AGENTIC ORCHESTRATION

LangGraph

WHAT IT IS

Stateful cyclic graph orchestration framework for multi-agent loops, state channels, and checkpoint memory.

HOW PRANAY APPLIES IT

Architected multi-agent customer support workflows with conditional intent routing across specialized agent nodes.

PRIMARY SYSTEM CONNECTION

CHAPTER 05 // ACTIVE FRONTIER

Currently Exploring & Prototyping

A clear demarcation between established verified competencies and the technical frontiers I am actively studying.

DEEPENING DOMAIN Active Research

LLM Engineering & Fine-Tuning

Studying Parameter-Efficient Fine-Tuning (PEFT, LoRA) using Hugging Face Transformers, quantization formats (GGUF, AWQ), and private inference runtimes.

AGENTIC SYSTEMS Prototyping

Advanced LangGraph & Memory Channels

Investigating complex cyclical agent graphs, dynamic human-in-the-loop validation checkpoints, and persistent cross-session memory channels.

DATA RETRIEVAL Benchmarking

Vector Databases & Hybrid RAG

Benchmarking vector similarity search (Chroma, Qdrant) alongside BM25 sparse keyword indices to build high-precision hybrid retrieval systems.

MULTIMODAL AI Exploration

Multimodal Video Understanding

Aligning audio speech markers with visual keyframe changes so AI agents can reason about technical presentations and code demonstrations on screen.

CHAPTER 06 // PROGRESSION

Engineering Journey

Chronological evolution showing growing system complexity — from procedural programming to distributed multi-agent systems and dataset platforms.

2024 • YEAR 1

Foundations • C, C++ & DSA

Began B.Tech in Computer Science and Engineering (Information Security) at VIT Vellore. Built deep foundations in memory allocation, pointers, Data Structures, and algorithmic complexity.

LATE 2024

Full-Stack & Security Certifications

Expanded into Python scripting, React web applications, and REST APIs. Earned verified industry credentials from IBM SkillsBuild (Cybersecurity) and Deloitte (Cyber Job Simulation).

EARLY 2025

Spendly • AI Personal Finance

Engineered Spendly using React, FastAPI, Firebase Firestore, and OpenRouter AI. Deployed live on Vercel with real-time balance tracking and predictive budgeting.

MID 2025

Sentinel AI • Cyber Threat Intelligence

Architected an asynchronous threat scanning pipeline integrating VirusTotal, WHOIS, and OSINT telemetry with Explainable AI correlation to streamline incident triage.

LATE 2025

SupportFlow AI • Multi-Agent Graphs

Constructed an agentic support platform using LangGraph, LangChain, SQLite state stores, and Ollama local LLMs to guarantee zero cloud data leakage.

EARLY 2026

Textora Engine • Video Dataset Platform

Built an open-source multimodal dataset engineering platform featuring caption-first dual ingestion, deterministic de-flickering, xxHash/SHA-256 deduplication, and SQLite WAL queues.

PRESENT • 3RD YEAR

10X Tech Internship & LLM Engineering

Contributing to AI/ML and software engineering at 10X Technologies with LangChain, Hugging Face, and RAG systems while maintaining an 8.27 CGPA at VIT Vellore.

CHAPTER 07 // VERIFIED CREDENTIALS

Academics & Profiles

B.TECH IN COMPUTER SCIENCE 2024 – 2028

Vellore Institute of Technology (VIT), Vellore

Specialization in Information Security

8.27 Current CGPA

3rd Year Academic Standing

Vellore, TN Campus Location

Relevant Core Coursework:

Data Structures & Algorithms, Object-Oriented Programming (Java), Database Management Systems (DBMS), Operating Systems, Computer Networks, Information Security, Discrete Mathematics.

GitHub • @Pranay-Kumar-02 LinkedIn • Pranay Kumar LeetCode • @Pranayyy_

Verified Industry Certifications

IBM SkillsBuild 2024

Cybersecurity Fundamentals

Core security principles, threat modeling, network defense architecture, and cryptography.

IBM Career 2024

Building Agents with Agentic AI

Multi-agent interaction patterns, tool calling, execution loops, and persistent agent state.

Deloitte 2024

Cyber Job Simulation

Practical threat triage, vulnerability auditing, incident response simulation, and log analysis.

Infosys Springboard 2024

Artificial Intelligence & Data Science

Machine learning workflows, neural network architectures, statistical modeling, and data pipelines.

IIT Madras (Techobytes & ChES) 2024

Cyber Ninjas with Ethical Hacking

Ethical hacking methodology, penetration testing concepts, vulnerability scanning, and defense.

CHAPTER 08 // COLLABORATION

Let's build something serious.

Actively open to Software Engineering internships, AI/ML research roles, and cybersecurity opportunities. Reach out directly for collaboration, interviews, or technical discussions.

Primary Email • Click to copy vonamala.pranay.official@gmail.com

Direct Phone +91 6301905015 LinkedIn linkedin.com/in/pranay-kumar-vonamala ↗ GitHub github.com/Pranay-Kumar-02 ↗

Send Direct Email Download Official Resume (PDF)