AI Agents Read Page
This page exposes the full portfolio in structured markdown for agents and crawlers.
# Khusham Bansal ## Role GSoC 2026 Contributor @ Meshery (CNCF) · Cloud-Native & AI Systems ## Summary Hi, I'm Khusham, a Computer Engineering student at Thapar Institute (CGPA 8.92) and a Google Summer of Code 2026 contributor at Meshery (CNCF/Layer5), working across AI reliability, distributed systems, and full-stack product engineering. I ground LLMs with knowledge graphs, build agentic systems and distributed infrastructure, and ship product surfaces that load fast. ## Location India ## Contact - Email: kbkhushambansal@gmail.com - Phone: +91-9464380089 - GitHub: https://github.com/KhushamBansal/ - LinkedIn: https://www.linkedin.com/in/khushambansal/ - LeetCode: https://leetcode.com/u/KhushamBansal/ - Resume: https://drive.google.com/file/d/1zNB0QHCn5YkywH_iOCTEaQg1nH5f88P3/view?usp=sharing ## Skills ### Languages - C/C++ - Go - Python - TypeScript - JavaScript ### AI / ML - PyTorch - TensorFlow - OpenCV - scikit-learn ### Web - React - Node.js - Express - Tailwind - Playwright ### Cloud & Data - AWS - Azure - Redis - MongoDB - Neo4j ## Experience ### Google Summer of Code Contributor — Layer5 / Meshery (CNCF) - Period: May 2026 — Present - Location: Remote - Selected under CNCF to develop Meshery, a cloud-native multi-mesh management platform for Kubernetes environments. - Authored 70+ PRs to Layer5 across UI reliability, registry workflows, navigation, Sistent integration, and Playwright E2E coverage. - Reviewed 130+ PRs across Layer5/Meshery. ### Open Source Contributor — Layer5 Ecosystem (Meshery, Kanvas, layer5.io) - Period: Dec 2025 — Apr 2026 - Location: Remote - Merged 18 PRs before GSoC selection (45 ecosystem-wide): Kanvas 11, meshery.io 4, layer5.io 3. - Shipped cursors, GitOps/collaboration features, program pages, and mobile improvements. ### Research Intern — AI Institute of the University of South Carolina (AIISC) - Period: Oct 2025 — Present - Location: Remote - Researching LLM reliability and techniques for reducing hallucinations. - Developing knowledge graph integration techniques to ground LLMs and improve factual accuracy. - Designing evaluation frameworks for accuracy and scalability with the Indian AI Research Organisation (IAIRO). ### Full-Stack Developer Intern — AltHire AI - Period: Oct 2024 — Sep 2025 - Location: Remote · Seattle, USA - Built landing pages for AltHire, MockMate, and TalentBoost. - Cut load times by 70% through optimized component architecture. - Reduced batch API calls from 7×n to ~10 total (98% less network overhead). - Automated Playwright E2E tests and added AWS SES alerts for interview-agent connection failures. ### Summer Intern — Thapar Institute of Engineering and Technology - Period: Jun 2025 — Jul 2025 - Location: Onsite · Patiala - Built DriveSafeAI — an IoT + multimodal AI system for real-time driver fatigue and stress monitoring. - Integrated computer vision, speech processing, and sensor pipelines for predictive safety signals. - Used ResNet, DenseNet, MobileNet for visual cues; Wav2Vec2, Whisper, and CNNs for audio classification. ### Computer Science Expert — Chegg - Period: Feb 2024 — Sep 2024 - Location: Online - Solved 150+ computer science questions for students worldwide. - Maintained a quality check score of 4.01/5. ## Open Source ### Layer5 / Meshery (CNCF) - Role: GSoC 2026 Fellow · 70+ PRs authored · 130+ reviewed - Period: Dec 2025 — Present - Contributing across the Layer5 ecosystem (Meshery, Kanvas, layer5.io) — UI reliability, registry workflows, navigation, Sistent integration, Playwright E2E coverage, GitOps/collaboration features, and program pages. 45+ PRs merged ecosystem-wide before GSoC selection. - Link: https://meshery.io/ ## Projects ### Ivo — Agent, RAG & MCP Personal Assistant - Description: Local personal AI agent with LangGraph orchestration (plan → tool calls → results), a FastAPI backend, a React cockpit, and SQLite + FTS5 RAG memory spanning semantic, episodic, and procedural recall with background consolidation. MCP gateway for Gmail, Calendar, and search, with human-in-the-loop approval on high-risk sends. - Period: 2026 - Stack: LangGraph, FastAPI, React, SQLite, MCP - Link: https://drive.google.com/file/d/16pp5bbNdyEUMzRb0Fj0CA-vsOgdfV82B/view?usp=sharing ### Milepost — ELD Trip Planner - Description: Plans property-carrying HGV trips and generates FMCSA driver daily logs under the 70hr/8-day HOS ruleset — routing, required fuel/pickup/drop-off stops, and one log sheet per calendar day in home-terminal time. - Period: 2026 - Stack: Django, DRF, React, TypeScript, PostgreSQL, Tailwind - Link: https://github.com/KhushamBansal/eld-trip-planner ### relayq — Distributed Job Queue - Description: Redis-backed durable job queue with leased workers, exponential-backoff retries, a dead-letter queue, and idempotent enqueue for horizontally scaled consumers. Drained 2,000 jobs at ~5.4k jobs/s with 4 workers; recovered in-flight jobs after a crash; 200 concurrent idempotent requests collapsed to 1 job. - Period: 2026 - Stack: Go, Redis - Link: https://github.com/KhushamBansal/relayq ## Blog Posts