CHENNAI, INDIA · CSE @ SSNCE

Build beyond
the ordinary.

I'm Vishwajith. I build systems that connect models, APIs, and infrastructure to solve real problems. Strong in machine learning, applied AI, and system design, I take projects from idea to reliable delivery.

SEE WHAT I'VE BUILT ↘GITHUB ↗LINKEDIN ↗

AI / ML
SOFTWARE ENGINEERINGOPEN TO
INTERNSHIPS

VISHWAJITH L K

VLK!

SCROLL TO EXPLORE↓

In brightest day, In blackest nightNo error shall escape my sight

ARTIFICIAL INTELLIGENCE✳SOFTWARE ENGINEERING✳MACHINE LEARNING✳ARTIFICIAL INTELLIGENCE✳SYSTEMS DESIGN✳SOFTWARE ENGINEERING✳MACHINE LEARNING✳ARTIFICIAL INTELLIGENCE✳

01THE BRIEF

A CLEAR SIGNAL THROUGH THE NOISE

I build systems where everything works together.

This means understanding how components connect—how AI models integrate with APIs, where data flows through a pipeline, how systems perform under load, and which trade-offs matter.

I believe in measuring model performance, building systems that scale, and making informed decisions about accuracy versus latency. Whether designing multi-agent orchestration, optimizing RAG systems, or architecting backend infrastructure, I focus on the complete system.

LET'S BUILD SOMETHING ↗

Vishwajith presenting a session on graph applications at SSN

SELECTED ACHIEVEMENTS2025—26

3^rd

GLOBAL PLACEMENT
MEDIAEVAL 2025

220+

LEETCODE PROBLEMS
SOLVED

₹2L

RELIANCE FOUNDATION
SCHOLARSHIP

8.607

CURRENT CGPA
AT SSNCE

15+

EVENTS ORGANISED
WITH THE TEAM

02SELECTED WORK

Selected work.

Filter by focus area.

AGENT-21 / MULTI-AGENT SIGNAL RESPONSE

LIVE
ORDERS

ANOMALY
ALERT

NEXT
ACTION

HACKSYMMETRIC · 202601 / 04

Agent-21

A multi-agent system that watches live delivery and revenue signals, makes sense of anomalies, and helps business owners respond.

Real-time event listeners and anomaly detection
PII-aware multilingual customer agent
Executive simulation for competing action plans

MULTI-AGENTREAL-TIMEBUSINESS AI

VIEW PROJECT ↗

OWNDB / DURABLE STORAGE VAULT

K/V

08 KBPAGEWAL

PERSONAL SYSTEMS · 202602 / 04

OwnDB

A disk-backed key-value store built from the ground up to understand durability below the application layer.

8 KB slotted pages and LRU buffer pool
B+ tree indexing for range scans
Checksummed WAL and recovery

STORAGE ENGINEB+ TREEWAL

VIEW SOURCE ↗

ENDOVISION / ENDOSCOPY IMAGE ANALYSIS

REGION
OF INTEREST

PALIGEMMA
2-3B

MEDIAEVAL · 202503 / 04

EndoVision

Fine-tuned PaliGemma 2-3B for GI endoscopy questions, taking the work from competition to published proceedings.

3rd place globally in Medico 2025 Task 1
Benchmarked with Accuracy, ROUGE-L, BLEU-4, BERTScore
Published in MediaEval 2025 workshop proceedings

COMPUTER VISIONRESEARCHMEDICAL AI

PUBLICATION RECORD ↗

DEEPFAKE DETECTION / AUTHENTICITY SHIELD

REAL

ALTERED

✓

FRAME ANALYSISVERIFYING

PEC HACKS 2.0 · 202404 / 04

Deepfake Detection

A real-time deepfake detection pipeline that kept working when the hackathon environment did not.

XceptionNet inference with quantization and pruning
OpenCV preprocessing and Django backend
MongoDB metadata with immutable logs

VIDEO AIOPENCVDJANGO

VIEW SOURCE ↗

03EXPERIENCE

EXPERIENCE / 2026

Built under
real conditions.

01NOW

2026REMOTE · INDIA

AI Software Engineering Intern

DriveThruData LLP

Worked on document intelligence and healthcare Q&A workflows. Evaluated TrOCR and Qwen2-VL; built across RAG, MCP, OCR, NER, context engineering, and AI workflow automation.

INTERNSHIP WORK ↗

04HACKATHONS