Arnav Malhotra

About

Bachelor of Engineering in Software Engineering

Expected May 2027 • Toronto, ON

Kuperman Investments • Remote

May 2026 – Present

Designed and deployed the firm's core market-data platform on AWS (S3/Parquet, DuckDB), creating a single source of truth over 3.2M+ records with automated freshness alerting
Engineered a fault-tolerant distributed collection system in Go (100+ concurrent workers) feeding real-time data into pricing and trading models
Eliminated silent data gaps with deduplication, automated backfill, and dead-letter handling, guaranteeing correctness under partial node failures

Quarkle.ai • Remote

Jul 2025 – Sep 2025

Cut median retrieval latency by 35% on a real-time AI document-Q&A pipeline by adding websocket streaming and request-level caching to the Python/FastAPI backend
Optimized a Next.js workspace for long-form documents with memoized rendering, batched state updates, and incremental loading, improving responsiveness by 40%
Grew weekly shared-document usage 2.1x across a 500+ user beta by shipping publishing and collaboration workflows with permission controls

Markis Investments • Remote

Apr 2025 – Jun 2025

Delivered an end-to-end resort-management suite (Next.js 15, React 19, TypeScript, Supabase/PostgreSQL) — a direct-booking website, operations dashboard, and guest portals — replacing manual spreadsheet workflows
Built the operations dashboard with month-over-month and year-over-year financial analytics (Chart.js), reservation and guest-profile management, and Hostaway API sync into PostgreSQL
Architected a multi-tenant QR-based event check-in system (7 role-based access tiers, 80+ API routes) that cut average guest processing from 2 minutes to under 20 seconds

York University • Toronto, ON

Oct 2024 – Apr 2025

Engineered a data scraping pipeline aggregating from 20+ sources with validation, retries, and exception handling, ensuring 98% accuracy in extracted datasets
Built a React-based campus events calendar adopted by 1,300+ students and faculty

Indian Institute of Technology, Roorkee • Roorkee, India

May 2024 – Jul 2024

Trained nnU-Net models on MRI data from 85 subjects with 5-fold cross-validation, achieving Dice score 0.52 ± 0.05 for epilepsy lesion segmentation; results published in Frontiers in Artificial Intelligence
Designed an automated slice-selection method ranking MRI slices by voxel intensity, improving lesion detection accuracy by 30% for FCD Type II

GoPythonJavaCTypeScriptJavaScriptSQL

AWS (EC2, S3, Lambda)gRPCRaftPrometheusDockerLinuxParquet

FastAPINext.jsReactNode.jsFlask

PostgreSQLDuckDBRedisMongoDBSQLite

PyTorchNumPyPandasPlaywrightGitTailwind CSS