Building high-concurrency, low-latency backend systems with a focus on performance optimization and scalable architecture.
💼 LinkedIn • 📧 Email • ⚔ Codeforces •
Backend-focused software engineer specializing in concurrent, low-latency systems across C++, Go, and Node.js. Passionate about systems architecture, optimizing execution hot-paths, and writing benchmark-verified code that handles real-world concurrent loads.
- Core: REST APIs, Concurrency & Multithreading, Low-Latency Design, Async Processing, Systems Programming
- Infrastructure & DBs: Redis, AWS (Lambda, DynamoDB, SQS, S3, API Gateway), MySQL, MongoDB
Defence Research and Development Organisation (DRDO) | Jan 2025 – Jun 2025
- 📉 Hot-Path Optimization: Cut encryption-test runtimes by 40% by optimizing C++ memory allocation and eliminating critical execution bottlenecks.
- 🛡️ Cryptanalysis: Identified cryptographic weaknesses by analyzing intermediate output patterns and tracing execution paths for statistical anomalies.
- 🤖 Test Automation: Built Python regression testing engines verifying correctness against 10,000+ known-answer vectors, reducing manual testing effort by 80%.
📊 Scalable Tie-Aware Leaderboard Engine (Go • Redis • C++)
- Engineered a custom bucket-index structure that reduced rank-update complexity from
$O(n \log n)$ to$O(n)$ , verified across 100K+ entries. - Achieved sub-millisecond p99 rank-read latency at 1,000 concurrent readers using Go atomic operations and goroutines for lock-minimal, thread-safe writes.
🛍️ MetaShop — AI-Driven Virtual Try-On Platform (Node.js • Redis • Diffusion Models)
- Built an asynchronous inference pipeline with a polling queue to decouple heavy diffusion models from UI rendering, ensuring a 97% job success rate under concurrent load.
- Integrated a Redis caching layer to eliminate redundant upstream calls, dropping average inference latency by 35% during burst traffic.
☁️ Serverless URL Shortener with Async Queue (AWS Lambda • DynamoDB • SQS • S3)
- Designed an event-driven SQS + Lambda pipeline with a dead-letter-queue (DLQ) retry mechanism, maintaining zero dropped events under 10x burst traffic load tests.
- Decoupled ingestion from heavy processing, enabling independent horizontal scaling of storage reads/writes.
🔐 FESLA — ML-Based AES Cryptanalysis (Python • SVM • MLP)
- Trained SVM and MLP classifiers on 30+ statistical ciphertext features, achieving 97% accuracy in detecting cryptographic leakage against a 10,000-sample test set.
- Problem Solving: Solved 900+ algorithmic problems across online judges.
- CodeChef: 5-Star Rating (Peak 1862)
- Codeforces: Expert
- Certifications: AWS Certified AI Practitioner (Feb 2026)
💡 "Performance isn't an optimization. It's a design decision."